Skip to main content

Overview

Welcome to the comprehensive guide for long-term forecasting and system modeling using iPool. This material covers structural modeling approaches, storage optimization, intelligent bidding, and cost-based dispatch.

1. Long Term Forecasting​

Long-term forecasting in iPool is used to simulate future electricity market conditions over extended periods such as months, years, or decades. The objective is not only to estimate prices, but also to evaluate system adequacy, market risks, generation expansion, storage operation, and long-term investment viability.

Unlike short-term forecasting, which focuses on immediate operational decisions and short-duration weather conditions, long-term forecasting focuses on broader structural and economic behavior of the power system.

iPool uses a structural simulation modeling approach, which means the actual power system structure, generation fleet, outages, fuel economics, storage behavior, transmission constraints, and market rules are modeled directly.

1.1 Forecasting Objectives​

Long-term forecasting generally has two primary objectives:

System Reliability​

System reliability studies focus on ensuring that the power system has enough available generation capacity to continuously supply electricity demand under both normal and abnormal conditions.

Because generators, transmission lines, and other electrical equipment can fail unexpectedly, the system must maintain reserve capacity capable of replacing unavailable generation during outages.

This reserve capacity is commonly referred to as system reserve.

Trading and Risk Management​

Trading and risk management studies focus on understanding future market prices, contract exposure, profitability, and operational strategies.

Examples include:

  • Evaluating whether a Battery Energy Storage System (BESS) investment is profitable
  • Testing generation bidding strategies
  • Assessing long-term contract viability
  • Forecasting future market prices
  • Studying renewable energy impacts on price volatility
  • Evaluating fuel and outage risks

These studies are typically more market-driven and price-sensitive than pure reliability studies.

1.2 System Reliability and Reserve Types​

Power systems require different types of reserve capacity to maintain stable operation during disturbances or outages.

Steady State / Dispatch Reserve​

Steady state reserve refers to generation capacity that can replace unavailable units during prolonged outages lasting several hours or days.

For example:

  • A coal unit unexpectedly trips offline
  • A generator enters forced outage
  • A transmission corridor becomes unavailable

The system must have enough alternative generation capacity available to continue supplying demand.

Spinning Reserve​

Spinning reserve refers to synchronized generating units that are already connected to the grid and capable of responding immediately to sudden disturbances.

Examples include:

  • Sudden generator trips
  • Circuit breaker operations
  • Transmission disturbances
  • Frequency instability events

Spinning reserve helps stabilize system frequency and is typically measured over seconds or minutes.

Traditional thermal generators contribute rotational inertia through turbine mass, while modern battery systems can provide extremely fast frequency response services.

1.3 Reliability Indices​

Several indices are used internationally to measure system adequacy and reliability.

LOLP (Loss of Load Probability)​

LOLP measures the expected number of hours where system demand may exceed available generation capacity.

For example:

  • Australia commonly targets approximately 2 hours per year of expected unserved energy risk.

LOLE (Loss of Load Expectation)​

LOLE measures the expected number of days in a year where there is a possibility of insufficient capacity to meet demand.

The Philippine DOE commonly uses LOLE as a planning reliability metric.

iPool can report both:

  • LOLP-based reliability metrics
  • LOLE-based reliability metrics
  • Expected unserved energy calculations

1.4 Structural Modeling Approach​

iPool primarily uses a structural modeling approach for long-term forecasting.

Structural modeling directly represents:

  • Power stations
  • Storage systems
  • Transmission behavior
  • Generator outages
  • Market bidding behavior
  • Fuel costs
  • Reserve requirements
  • Demand growth
  • Renewable energy variability

This approach is significantly more detailed than purely statistical or economic forecasting models.

Alternative forecasting approaches include:

  • Statistical forecasting
  • Econometric forecasting
  • Hybrid forecasting
  • Game theory approaches
  • Machine learning forecasting

However, structural simulation provides stronger transparency and operational realism for electricity market studies.

1.5 Forecasting Challenges​

Electricity markets are inherently difficult to forecast because electricity cannot be economically stored at large scale without dedicated storage infrastructure.

Major forecasting challenges include:

Non-Storability of Electricity​

Unlike physical commodities such as rice, oil, or metals, electricity must generally be generated and consumed simultaneously.

This creates strong price sensitivity during:

  • High demand periods
  • Generator outages
  • Fuel shortages
  • Renewable energy variability

Transmission Constraints​

Power systems contain physical network limitations.

Examples include:

  • Thermal transmission limits
  • Congestion constraints
  • Voltage stability constraints
  • Frequency stability limits

Transmission lines have thermal current limits. Excessive current causes line heating, which can damage equipment or trigger protective shutdowns.

High-voltage transmission systems are used primarily to reduce current flow and minimize thermal losses.

Stochastic Generator Availability​

Generator outages are stochastic, meaning they are random and probabilistic in nature.

iPool models generator behavior using probability distributions such as:

  • Weibull distributions
  • Forced outage rates
  • Mean time to fail
  • Mean time to repair

Because future outages cannot be predicted exactly, long-term forecasting always involves uncertainty.

Variable Renewable Energy (VRE)​

Variable Renewable Energy sources such as:

  • Solar
  • Wind
  • Biomass
  • Geothermal

are highly intermittent and weather-dependent.

Wind forecasting is particularly difficult because wind output can change rapidly over short time intervals.

iPool supports historical trace-based renewable forecasting using:

  • MRH CFAT hourly trace files
  • Historical generation traces
  • Site-based renewable modeling

AI-assisted renewable forecasting tools may also be used for short-term renewable forecasting applications.

1.6 Demand Forecasting Considerations​

Demand forecasting determines future electricity consumption patterns and load behavior.

Key considerations include:

Daily Load Shape​

Demand shape changes depending on:

  • Dry season
  • Rainy season
  • Summer
  • Winter
  • Spring
  • Fall

Different seasons create different consumer behavior and load characteristics.

Day Types​

Different day categories produce different load profiles:

  • Workdays
  • Saturdays
  • Sundays
  • Holidays
  • School days
  • Shopping days

iPool commonly models:

  • Weekdays
  • Saturdays
  • Sundays/Holidays

using separate typical demand profiles.

Demand Growth​

Long-term studies typically assume future:

  • Energy growth
  • Peak demand growth
  • Economic growth

These assumptions are often sourced from economic planning studies or utility forecasts.

Load Factor and Peakiness​

Load factor describes whether system demand is:

  • Flat and stable
  • Thin and peaky

Peaky systems are more expensive because they require peaking plants such as:

  • Diesel
  • Oil
  • Gas turbines

Base load plants operate most efficiently when running continuously at stable output levels.

Battery systems are increasingly important for reducing short-duration demand peaks.

1.7 Generation Forecasting Considerations​

Generation forecasting is generally more difficult than demand forecasting because generation availability and bidding behavior are highly uncertain.

Cost-Based vs Bid-Based Modeling​

Cost-Based Dispatch​

Cost-based dispatch uses generator operating costs to determine dispatch order.

This method:

  • Is faster
  • Produces stable prices
  • Is suitable for reliability studies
  • Uses Economic Lambda Dispatch principles
Bid-Based Dispatch​

Bid-based dispatch simulates actual market bidding behavior.

This approach is affected by:

  • Contract positions
  • Market strategies
  • Fuel prices
  • Time-of-day economics
  • Day-type behavior
  • Storage strategies

Bid-based modeling is more realistic for market price forecasting but significantly more complex.

Availability Modeling​

Generation availability consists of:

Planned Outages (POR)​

Planned outages represent scheduled maintenance events.

Examples include:

  • Annual maintenance
  • Major overhauls
  • Inspection shutdowns

Maintenance is normally scheduled outside peak demand periods.

Forced Outage Rate (FOR)​

Forced outages are unexpected generator failures.

FOR measures the percentage of time a generator becomes unavailable unexpectedly.

Typical thermal plant availability may range from:

  • 70% to 95%

Well-maintained coal plants commonly target:

  • 85% to 90% availability

Energy-Limited Plants and Storage​

Some generators are energy-limited rather than capacity-limited.

Examples include:

  • Pumped storage hydro
  • Reservoir hydro
  • Battery Energy Storage Systems (BESS)

Hydro availability depends on:

  • Rainfall
  • Reservoir inflows
  • Irrigation policies
  • Seasonal conditions

Battery operation depends on:

  • Charging strategies
  • Market prices
  • Round-trip efficiency
  • State of charge constraints

Variable Renewable Energy and Consumer Solar​

Renewable energy growth significantly changes system demand behavior.

Consumer solar adoption reduces daytime grid demand, creating the well-known Duck Curve effect.

This can:

  • Reduce daytime prices
  • Create steep evening ramps
  • Reduce baseload plant profitability
  • Increase storage requirements

iPool models VRE using historical renewable traces and technology-specific generation profiles.

Typical VRE technologies include:

  • Solar
  • Wind
  • Biomass
  • Geothermal

Each renewable site may require separate trace modeling and capacity assumptions for future forecast years.

Typical Bid Aggregation​

Long-term bid forecasting commonly uses:

  • Monthly bid aggregation
  • Day-type bid aggregation
  • Seasonal bid structures

Typical categories include:

  • Weekday
  • Saturday
  • Sunday/Holiday

Hydro and energy-limited plants may require additional monthly adjustments based on historical capacity factors and water availability assumptions.

Model Validation and Assumptions​

All forecasting models must be validated against historical market behavior.

Common validation methods include:

  • Historical replay simulations
  • Backcasting
  • Price comparison studies
  • Dispatch comparison studies

Because future conditions are uncertain, all long-term forecasts rely heavily on assumptions.

Examples include:

  • Fuel price assumptions
  • Demand growth assumptions
  • Renewable growth assumptions
  • Future generator entries
  • El Niño and La Niña conditions
  • Transmission developments
  • Policy changes