Chapter 1: Introduction

Overview of Power System Operation

Electric power systems are large, complex networks designed to supply electricity to
consumers reliably and economically. They include generation plants, transmission lines,
substations, transformers, and distribution systems. Operating such a system requires
constantly balancing electricity generation with demand, while keeping the system secure,
stable, and efficient.
Power system operation can generally be grouped into three levels:
• Long-term planning, where future generation and transmission expansions are
planned.
• Short-term planning, which involves scheduling generation units, fuel planning, and
maintenance decisions.
• Real-time operation, where load-frequency control, economic dispatch, and power
flow monitoring occur.
Among these tasks, Unit Commitment (UC) plays a vital role because it determines which
power plants should run at what time to meet the forecasted load at minimum cost.

1.2 Role of Unit Commitment (UC)

Unit Commitment is the decision-making process of planning which generating units to turn
ON or OFF over a specified period—typically 24 hours. It also decides how much power each
committed unit should generate.
UC must ensure:
• The total generation meets the load at all times
• Units operate within their technical limits
• System reliability and spinning reserve requirements are satisfied
• The overall cost of operation is minimized
UC is challenging because it combines binary decisions (ON/OFF status of units) with
continuous decisions (power output), making it a mixed-integer optimization problem. It
also must incorporate physical constraints such as minimum up/down time, ramping limits,
start-up and shut-down costs, and reserve requirements.
Due to these complexities, UC is one of the most computationally intensive problems in power
system optimization.

1.1Need for Deterministic UC

In a deterministic UC model:
• The load forecast is treated as exact
• Generator availability is assumed to be known
• Renewable generation and uncertainties are ignored
Although modern grids face significant uncertainty due to renewables, deterministic UC still
plays an important role, especially because:
1. It forms the foundation for advanced models such as stochastic or robust UC.
2. It is simpler and faster to compute, making it suitable for many traditional grids.
3. It is widely used in academic research and as a benchmark for testing algorithms.
4. Many utilities with low renewable penetration still rely on deterministic scheduling.

1
1.4 Importance of Transmission Constraints
Earlier UC approaches assumed that the entire power system could be treated as a single “bus,”
ignoring the transmission network. However, with increasing power demand, renewable
integration, and congestion issues, ignoring transmission constraints can lead to unrealistic
or unsafe schedules.
Including transmission constraints helps ensure:
• Line flow limits are respected
• System voltage and stability remain within limits
• Thermal ratings of lines and transformers are not exceeded
• Congestion is properly managed
• Power flows remain within safe operating conditions
Incorporating these constraints transforms the UC problem into Security-Constrained Unit
Commitment (SCUC). This provides results that are more realistic and usable for real-world
operations. Without transmission constraints, committed units may seem optimal
mathematically but may cause overloads or violations when applied to the physical grid.

1.5 Introduction to MATPOWER and MOST

MATPOWER is a widely used, MATLAB-based power system simulation tool. It provides:
• Power flow (AC and DC) solvers
• Optimal Power Flow (OPF) modules
• Data input formats for buses, branches, and generators
• A flexible environment for academic and research purposes
MOST (MATPOWER Optimal Scheduling Tool) builds on MATPOWER and adds
advanced capabilities for time-coupled optimization problems such as:
• Deterministic Unit Commitment
• Multi-period Optimal Power Flow
• Security-Constrained UC
• Energy storage scheduling
• Reserve management
The example script most_ex6_uc is a complete demonstration of deterministic UC with
transmission constraints in MATLAB. This makes MATPOWER + MOST a powerful
combination for both learning and real-world research applications.

1.6 Objectives of the Study

The main objectives of this documentation are:
1. To understand the concept of deterministic Unit Commitment and its importance in
daily power system operation.
2. To study how transmission constraints influence power flows, generator schedules, and
system cost.
3. To implement a UC case using the MATPOWER/MOST environment.
4. To examine the structure and role of input data files (bus, branch, and generator tables).
5. To analyze the OPF and UC results produced by MOST, including LMPs, commitment
status, and cost distribution.
6. To create a clear, structured report that can serve as a reference for students and
researchers.

2
 Chapter 2: Literature Review

2.1 Classical UC Approaches

The Unit Commitment problem has been studied for several decades, and some of the earliest
solutions were built around simplifying assumptions to make the problem easier to compute.
Traditional approaches mainly included:
• Priority-List Techniques – Generators were arranged in order of their cost or
efficiency, and decisions were made using rule-based checks. These methods were
simple and fast but could not capture realistic operating constraints like ramp rates or
start-up costs.
• Dynamic Programming (DP) – One of the first systematic methods used for UC. DP
could represent key constraints such as minimum up/down time, but its computational
burden increased dramatically as the number of units grew, limiting scalability.
• Lagrangian Relaxation (LR) – A breakthrough technique that decomposed the UC
problem into smaller independent subproblems. LR significantly reduced computation
time but often produced “near-feasible” results that required additional adjustments.
These classical approaches helped shape the field but lacked the ability to handle the
complexities of modern power systems, especially those with network congestion and
renewable energy variability.

2.2 Mathematical Programming Methods

As computing power improved, more advanced mathematical optimization techniques became

the preferred approach for solving UC.
Mixed-Integer Linear Programming (MILP)
MILP-based formulations became popular because they provide:
• A clean, structured mathematical model
• The ability to represent most practical power system constraints
• Reliable optimal solutions for linearized problems
With modern solvers such as Gurobi and CPLEX, MILP can handle realistic UC models that
include ramp limits, reserve margins, start-up costs, and commitment logic.
Mixed-Integer Nonlinear Programming (MINLP)
Some UC applications require handling non-linear cost curves or AC power flow equations.
MINLP models are suitable in such cases but are more difficult to solve due to the nonlinearity
and non-convexity involved.
Security-Constrained UC (SCUC)
SCUC enriches the UC formulation by integrating the transmission network directly into the
optimization problem. This approach captures the impact of congestion, line limits, and
network security requirements, making it essential for real-world electricity markets.
Mathematical programming models offer the accuracy and flexibility needed for modern power
system planning, though they can still face computational challenges for very large systems.

3
2.3 Transmission-Constrained UC in Research
Research on UC with transmission constraints has expanded significantly as power systems
deal with higher loads, stressed networks, and increasing renewable penetration. Studies in this
area typically emphasize:
• DC-based transmission models, which simplify the network but still capture line
flows and congestion.
• AC-based SCUC, which is more detailed but computationally heavier due to nonlinear
voltage equations.
• Congestion-aware scheduling, where commitment decisions actively consider line
thermal limits and transformer capabilities.
• Locational Marginal Pricing (LMP), which depends directly on transmission
constraints and provides location-specific energy prices.
Many studies highlight that ignoring transmission effects can produce schedules that are
impossible—or unsafe—to operate. Because of this, transmission-constrained UC is now
standard practice in most system operators and energy markets. However, scaling these models
for large systems remains a challenge in research.

2.4 Tools and Platforms for UC: MATPOWER vs MOST vs Commercial

Solvers
Several software platforms have been developed to support UC analysis:
MATPOWER
MATPOWER is a widely used open-source MATLAB toolbox that provides:
• AC and DC load flow analysis
• OPF solvers
• Easy-to-use data structures for generators, buses, and branches
However, MATPOWER alone does not include UC functionality.
MOST (MATPOWER Optimal Scheduling Tool)
MOST extends MATPOWER and enables:
• Multi-period UC simulations
• Modeling of start-up/shut-down costs
• Ramp rate limits
• Reserve handling
• Transmission-constrained scheduling
MOST is ideal for academic research due to its transparency and flexibility.
Commercial Tools (e.g., Gurobi, CPLEX, PLEXOS)
Commercial solvers offer:
• Highly optimized MILP and MINLP engines
• Advanced SCUC models
• Faster execution for large-scale networks
The drawback is the high cost and closed-source nature, making them less suitable for open
academic experimentation.
In comparison, MATPOWER + MOST provides a practical, accessible framework for research
and teaching, even if commercial solvers remain faster for large industrial applications.

4
2.5 Gaps in the Literature
Despite extensive research on UC, several important gaps still exist:
1. Limited open-source SCUC tools
Many detailed models remain proprietary, creating a barrier for academic and research
communities.
2. Insufficient treatment of uncertainty
Deterministic UC overlooks the variability in loads and renewable energy, making its
assumptions less valid for modern grids.
3. Computational complexity
High-fidelity SCUC models involving non-linear AC power flow remain difficult to
solve at scale.
4. Integration of emerging technologies
Distributed energy resources, electric vehicles, and large-scale storage systems are not
fully addressed in many existing UC frameworks.
5. Lack of standard benchmark systems
Comparing algorithms fairly is challenging because few publicly available UC datasets
include detailed transmission networks.
Addressing these gaps would support the development of more accurate, scalable, and future-
ready UC models.

5
 Chapter 3: Fundamentals of Unit Commitment
3.1 Definition and Purpose
Unit Commitment (UC) is a core decision-making process in power system operation. Its main
purpose is to determine which generating units should be turned ON or OFF over a given
time horizon—typically 24 hours or more—and how much power these generating units should
produce during each period.
Unlike real-time dispatch, which focuses only on the next few minutes, UC plans generator
schedules well ahead of time to ensure:
• The forecasted demand is met at every hour
• System reliability is maintained with adequate reserve
• Operating cost is minimized
• Technical limits of generators and the network are respected
UC helps system operators avoid unnecessary start-ups, maintain efficient operation, and
prevent system failures. Without UC, utilities would face unpredictable costs, inefficient
generator usage, and potential reliability issues. Thus, UC is the backbone of day-ahead
planning in modern power systems.

3.2 UC Scheduling Horizon and Time Resolution

A UC problem is always defined over a planning horizon, which is divided into discrete time
intervals.
Scheduling Horizon
The horizon typically varies from:
• 24 hours (most common)
• 48 hours for complex systems
• One week for systems with hydro scheduling requirements
A longer horizon gives the operator a better view of future demand and allows more
coordinated planning, especially when start-up times are long.
Time Resolution
Time is split into intervals (time steps). Common resolutions include:
• Hourly (standard in most UC models)
• 30-minute intervals (used in some countries and markets)
• 15 minutes for systems with high renewable penetration
A smaller time step gives more accurate modeling of ramping and demand variability but
increases the computational burden.
Choosing the right horizon and resolution is a trade-off between accuracy and computational
complexity. For teaching and research, hourly UC over 24 hours is the standard.

3.3 Binary Commitment Variables

The unique challenge in UC arises from the existence of binary variables. For every generator
i and time t, the model defines a binary variable:
1 if generator 𝑖 is ON at time 𝑡
𝑢𝑖,𝑡 = {
0 if generator 𝑖 is OFF at time 𝑡

6
These binary decisions make UC a Mixed-Integer Programming (MIP) problem. When a
unit is ON (u = 1), the model determines how much power it should produce. When the unit is
OFF (u = 0), the output is set to zero.
Binary variables also link operating constraints such as:
• Minimum up and down times
• Start-up and shut-down logic
• Ramp rate enforcement
• Reserve allocation
Because binary variables significantly increase computational complexity, UC is much harder
to solve than standard optimal power flow (OPF), which deals only with continuous variables.

3.4 Economic Dispatch vs Unit Commitment

Understanding the difference between Economic Dispatch (ED) and Unit Commitment is
essential.
Economic Dispatch (ED)
Determines how much power each committed generator should produce at a specific instant,
assuming the ON/OFF status of generators is already known. ED deals only with:
• Continuous power outputs
• Minimizing cost at a single time step
• Meeting demand and reserve constraints
ED is a simpler optimization problem solved every few minutes in real-time operation.
Unit Commitment (UC)
UC operates on a multi-hour horizon. It decides:
• Which units should run, and when
• When units should start or stop
• How generator output should evolve over time
UC incorporates long-term operational constraints like minimum up/down time and start-up
costs, which ED cannot handle.
In simple terms:
• UC is the planning layer (day-ahead)
• ED is the operational layer (real-time)
UC includes ED as a subproblem, but ED does not include UC.

3.5 Unit Commitment Constraints

UC must respect various technical and operational constraints. These constraints ensure that
generator schedules are realistic and feasible in actual system operation.

3.5.1 Minimum Up/Down Time

Every thermal generator has physical limits governing how long it must remain ON or OFF
after a switching event.

Minimum Up Time

Once a unit is turned ON, it must stay ON for a minimum number of hours before it can be
shut down.

7
Example:
If a unit has a minimum up time of 4 hours, it cannot turn off immediately after being started.

Minimum Down Time

Once a unit is turned OFF, the machine must cool and stabilize before it can restart.
Example:
A generator with a 3-hour minimum down time cannot restart within 3 hours of shut-down.
These constraints prevent excessive cycling, reduce mechanical wear, and ensure safe
operation of the unit. They also significantly impact scheduling flexibility, making the UC
problem more complex.

3.5.2 Ramp Rate Limits

Ramp rate limits define how quickly a generator can increase or decrease its power output.
There are generally two types:
• Ramp-up rate: Maximum increase in power per hour
• Ramp-down rate: Maximum decrease in power per hour
Ramp constraints are crucial when load changes rapidly, especially with renewable variability.
A generator cannot instantly jump from minimum to maximum output—it must follow its
ramping profile.
Ramp limits often interact with commitment decisions. For example, a unit might be forced to
remain ON longer just to support a required ramping need in the next interval.

3.5.3 Start-Up and Shut-Down Costs

Thermal power plants incur additional costs when starting or stopping because of:
• Fuel needed for pre-heating
• Mechanical stress on equipment
• Labor and operational expenses
Start-up costs may vary depending on how long the unit has been offline (“cold start,” “warm
start,” “hot start”). Shut-down costs are generally smaller but still relevant.
These costs significantly influence UC decisions. Sometimes, it is cheaper to keep a generator
running at low output than to shut it down and restart it later.

3.5.4 Spinning Reserve Requirements

Power systems must maintain a margin of flexible capacity known as spinning reserve to
respond instantly to sudden changes such as:
• Unexpected load increase
• Generator failure
• Renewable variability
UC ensures enough units are ON and loaded below their maximum capacity to provide the
required spinning reserve. This sometimes forces units to operate even if they are not
economically optimal, prioritizing reliability over cost.
Reserve constraints are particularly critical in systems with high renewable penetration.

8
3.6 Objective Function: Minimizing Total Operating Cost
The primary goal of UC is cost minimization. The objective function generally includes:
1. Fuel costs
2. Start-up and shut-down costs
3. No-load costs (cost of keeping a unit ON even at zero output)
4. Reserve provision costs
Mathematically, UC minimizes:
Total Cost = ∑(Fuel Cost𝑖,𝑡 + No-Load Cost𝑖,𝑡 + Start-Up𝑖,𝑡 + Shut-Down𝑖,𝑡 )
𝑖,𝑡

Subject to all operational and power balance constraints.

This ensures the system is operated:
• Economically
• Reliably
• Within technical limits
The final UC schedule is therefore a compromise between cost efficiency and system
reliability.

9
 Chapter 4: Transmission Constraints in Unit Commitment
Transmission constraints play a central role in shaping modern Unit Commitment (UC)
decisions. As power systems operate closer to their limits and renewable generation increases,
the network can no longer be treated as an “unconstrained pipe.” The transmission grid directly
influences which generators can run, how much they can produce, and whether their power can
safely reach the load. This chapter explains the key concepts behind transmission modeling in
UC, focusing on practical tools like the DC power flow, PTDFs, congestion management,
Locational Marginal Pricing (LMP), and the broader framework of Security-Constrained
Unit Commitment (SCUC).

4.1 DC Power Flow Model

In most transmission-constrained UC studies, the DC power flow model is used as the network
representation. While the full AC power flow model is more accurate, it is also nonlinear and
significantly harder to incorporate into large-scale optimization problems. The DC
approximation strikes a balance by simplifying the physics of power flow while still capturing:
• Real power transfer between buses
• Line loading and congestion
• Impact of generator dispatch on network flows

Assumptions Behind DC Power Flow

The DC model relies on several assumptions:

1. Voltage magnitudes are close to 1.0 pu, so their variation has minimal impact on real-
power flow.
2. Voltage angles are small, allowing linear approximations for trigonometric terms.
3. Reactive power and losses are ignored.
4. Line resistances are much smaller than reactances, meaning power flow is mostly
determined by line reactance.
These simplifications turn the AC equations into a linear relationship:
𝜃𝑖 − 𝜃𝑗
𝑃𝑖𝑗 =
𝑋𝑖𝑗

where
𝑃𝑖𝑗 = real power flowing from bus i to bus j
𝜃𝑖 and 𝜃𝑗 = bus voltage angles
𝑋𝑖𝑗 = line reactance

Why DC Power Flow Is Used in UC

• It keeps the optimization problem linear, enabling large-scale MILP solvers.

• It is sufficiently accurate for representing congestion and line overloads, especially in
high-voltage transmission networks.
• It computationally scales well to multi-period problems like UC.
Even though it is a simplification, the DC power flow model remains the standard network
model in UC research and market clearing algorithms around the world.

10
4.2 Power Transfer Distribution Factors (PTDF)
When adding transmission constraints to the UC problem, we need a fast way to determine
how generator injections affect line flows. This is where Power Transfer Distribution
Factors (PTDFs) come in.

What PTDFs Represent

A PTDF indicates how much power on a specific line will change when 1 MW is injected at
one bus and withdrawn at another. For example:
• A PTDF of 0.3 means that a 1 MW injection will create a 0.3 MW flow on that line.
• A negative PTDF means the flow moves in the opposite direction.

Role of PTDFs in UC

PTDFs help translate generator outputs into line flow constraints. Instead of solving separate
power flow equations at every step, UC models use PTDF matrices to compute flows directly:
𝐹ℓ,𝑡 = ∑ 𝑃𝑇𝐷𝐹ℓ,𝑔 ⋅ 𝑃𝑔,𝑡
𝑔

where
𝐹ℓ,𝑡 = flow on line ℓat time 𝑡
𝑃𝑇𝐷𝐹ℓ,𝑔 = sensitivity of line flow to generator g
𝑃𝑔,𝑡 = generator output

Advantages of PTDF-Based Modeling

•Eliminates the need for iterative solvers inside the optimization process
•Keeps the problem linear and scalable
•Allows congestion effects to be incorporated efficiently
•Enhances transparency and interpretability of results
PTDF-based SCUC formulations are commonly used by ISOs such as PJM, MISO, and
ERCOT.

4.3 Thermal Line Limits

Every transmission line has a thermal capacity, which is the maximum power it can carry
without overheating. When a line carries too much power, the conductor temperature increases,
which can lead to:
• Sagging lines
• Damage to insulation
• Reduced lifetime of the conductor
• Potential safety hazards

11
Types of Line Limits

1. Seasonal Limits – e.g., higher limits in winter due to cooler air

2. Emergency Limits – short-term rating for contingencies
3. Normal Operating Limits – day-to-day operating thresholds
4. Dynamic Line Ratings – adaptive limits based on real-time weather

When a line approaches its limit, the UC model must adjust generator schedules to prevent
overload.

Mathematical Representation in UC

For each line ℓand time period 𝑡:

−𝐹ℓ𝑚𝑎𝑥 ≤ 𝐹ℓ,𝑡 ≤ 𝐹ℓ𝑚𝑎𝑥

These constraints often drive decisions such as:

• Turning on a generator near the load center
• Reducing output from generators located far away
• Changing the power flow path through redispatch
• Triggering congestion-based pricing (LMP differences)

4.4 Network Congestion and Locational Marginal Pricing (LMP)

Congestion occurs when a transmission line reaches its limit and prevents the system from
using the lowest-cost generators. When this happens, the solution shifts to higher-cost
generators located near the constrained area.

How Congestion Creates LMP Differences

LMP is defined as the cost of supplying 1 additional MW of power at a specific bus. When no
congestion exists, LMPs are equal across the system because power can flow freely.
But when a line is congested, LMPs diverge:
• Higher LMP at the load side of the congested line
• Lower LMP at the generation side
This price separation reflects the increased cost of delivering energy to that location.
Importance of LMP in UC
• Encourages generators in the constrained area to increase output
• Signals the true cost of serving load at each bus
• Guides investment in transmission upgrades
• Provides incentives for demand response
Congestion and LMP are deeply interconnected with UC decisions because generator
commitment depends not just on marginal cost, but on locational feasibility.

12
4.5 Impact of Transmission Constraints on UC Scheduling
When transmission constraints are ignored, UC tends to schedule low-cost generators even if
they are far from major loads. But once transmission is considered, the entire dispatch pattern
can change.

Key Impacts

1. Altered Generator Commitment Decisions

High-cost but well-located generators may start earlier or run longer because cheap generators
cannot deliver power past congested lines.

2. Increased Operating Costs

Transmission limits force the system to use more expensive generation, increasing total UC
cost.

3. Local Reliability Requirements

Some regions might require local generation to stay online to maintain voltage stability or meet
N-1 security criteria.

4. Curtailment of Renewable Energy

Wind or solar plants in remote locations may be curtailed if transmission is insufficient.

5. Multi-period Effects

Congestion in one hour affects future hours because UC decisions include start-up times and
minimum up/down constraints.

Example Scenario

A wind farm located far away might produce cheap energy. However, if the line connecting it
to the grid is congested:

• The wind farm’s output may be reduced

• A local thermal generator may be started
• System cost increases
• LMP at the load center rises

This demonstrates why transmission-aware UC is essential for realistic system planning.

13
4.6 Security-Constrained Unit Commitment (SCUC) Concepts
SCUC extends the traditional UC model by incorporating network constraints, contingency
analysis, and security rules. It is the standard tool used by Independent System Operators
(ISOs) worldwide.

Core Features of SCUC

1. Network-Constrained Dispatch

Incorporates DC power flow, PTDFs, thermal limits, and congestion.

2. N-1 Contingency Analysis

The system must remain secure even after the loss of any single component:
o Generator outage
o Transmission line outage
o Transformer failure

3. Reserve Requirements

Ensures adequate spinning and non-spinning reserves for emergencies.

4. Commitment Logic

Maintains minimum up/down time, ramp rates, and start-up dynamics.

5. Locational Pricing Integration

SCUC is used to set LMPs in electricity markets.

Why SCUC Is Essential

• Prevents blackouts by ensuring secure operation

• Accounts for real-world limits on power flows
• Ensures fairness and transparency in market operations
• Helps maintain system stability despite renewable variability
• Produces dispatch solutions that are physically feasible and safe

SCUC models are computationally demanding but necessary for ensuring both economic
efficiency and system reliability.

14
 Chapter 5: Overview of MATPOWER and MOST

5.1 About MATPOWER

MATPOWER is a MATLAB-based open-source toolbox widely used in power system
simulation and optimization. It was originally developed to support research and teaching,
offering a transparent, flexible, and modifiable framework for various power system
studies.
Its primary goal is to help users analyze steady-state system behavior and perform economic
optimization in a way that is accessible, yet sufficiently sophisticated for academic research.
MATPOWER has grown into a standard tool in the field, largely because it balances ease of
use with robust functionality.

Key Advantages of MATPOWER

1. Ease of Use

Users can write MATLAB scripts that are simple to understand and modify. Its
functions are modular, making it easy to adapt for different case studies.

2. Flexibility

MATPOWER allows the addition of new features or constraints, enabling researchers

to experiment with innovative optimization techniques without needing commercial
software.

3. Open Source

The source code is freely available. Researchers can analyze, customize, or extend any
part of the code to suit their requirements.

4. Efficiency and Accuracy

MATPOWER supports both AC and DC power flow solutions, providing accurate

representation of voltage, flows, and losses in transmission networks.

5. Integration-Friendly

Its design allows integration with tools like MOST for multi-period optimization and
unit commitment, making it ideal for academic research on real-world scheduling
problems.

MATPOWER is therefore an ideal platform for both learning and research, bridging the gap
between classroom studies and industrial applications.

15
5.2 Power Flow & Optimal Power Flow Capabilities
MATPOWER provides two fundamental functionalities: Power Flow (PF) and Optimal
Power Flow (OPF). Together, these form the foundation of deterministic unit commitment
studies.

5.2.1 Power Flow Analysis

Power flow analysis determines the steady-state operating condition of a power system for a
given set of generator outputs and load demands. Key outputs include:
• Voltage magnitude and phase at each bus
• Real and reactive power flows in transmission lines
• Generator outputs and bus injections

MATPOWER supports two types of power flow:

1. AC Power Flow
o Models both real and reactive power
o Accounts for voltage magnitude and angle variations
o Nonlinear in nature, requires iterative methods for solution
o Suitable for detailed studies of voltage stability and reactive power management
2. DC Power Flow
o Linear approximation focusing on real power only
o Ignores reactive power and voltage magnitude variations
o Uses bus voltage angles to calculate line flows
o Computationally efficient, ideal for multi-period UC and transmission-
constrained optimization

DC power flow is particularly useful for large-scale UC studies because it allows network
constraints to be incorporated without introducing nonlinearity, keeping the problem
solvable using MILP techniques.

5.2.2 Optimal Power Flow (OPF)

OPF builds on power flow by introducing an optimization objective. In MATPOWER, OPF

can minimize:

• Fuel costs of generators

• System losses
• Operating cost while respecting network limits

OPF also allows inclusion of constraints like:

• Maximum and minimum generator outputs

• Transmission line limits
• Voltage magnitude bounds
The linear DC OPF is widely used in UC because it balances solution accuracy with
computational efficiency. AC OPF is used when voltage or reactive power issues are critical.

16
5.3 Introduction to MOST (MATPOWER Optimal Scheduling Tool)
While MATPOWER provides excellent tools for OPF, it does not handle multi-period unit
commitment problems directly. The MATPOWER Optimal Scheduling Tool (MOST)
extends MATPOWER to address this limitation.
MOST integrates:

• Unit commitment logic

• Start-up and shut-down cost modeling
• Ramping and minimum up/down constraints
• Transmission-constrained optimization
• Reserve scheduling

This makes MOST a powerful tool for day-ahead scheduling and deterministic UC studies.

Applications of MOST
1. Deterministic UC
MOST can simulate day-ahead commitment decisions for all thermal and renewable
generators in the system.
2. Security-Constrained UC (SCUC)
Using DC OPF constraints, MOST can ensure schedules respect line limits and
contingency requirements.
3. Reserve Allocation
MOST can incorporate spinning and non-spinning reserve requirements for system
reliability.
4. Market Studies
Researchers can use MOST to study locational marginal pricing (LMP), congestion
costs, and generator profits under realistic network constraints.

By combining MATPOWER and MOST, researchers can perform advanced UC studies

without relying on expensive commercial solvers like PLEXOS or Gurobi, while still
producing realistic and reliable results.

5.4 MOST Data Structure and Input Files

To run MOST, the power system must be defined through structured input files. These files
provide the network, generator, and load information in a standardized format.

5.4.1 Bus Data

The bus file defines the nodes of the network:

• Bus number and type (load, generator, slack)
• Real and reactive power demand
• Voltage limits
• Shunt elements
This data allows MOST to calculate power injections and enforce network constraints.

17
5.4.2 Branch Data

Branch files describe transmission lines and transformers:

• From and to buses
• Line resistance and reactance
• Thermal limits (maximum power flow)
• Transformer tap ratios
Branch data ensures that power flows respect thermal limits and are feasible across the
network.

5.4.3 Generator Data

Generator files contain:

• Maximum and minimum generation limits
• Fuel and operating costs
• Ramp-up and ramp-down limits
• Minimum up and down times
• Start-up and shut-down costs
This data is critical for MOST to schedule generators optimally while adhering to operational
constraints.

5.4.4 Optional Input Files

• Forecasted load profiles

• Renewable generation scenarios
• Reserve requirements
These additional inputs allow for realistic, multi-period scheduling and reliability studies.

5.5 Features of MOST for Unit Commitment

MOST extends MATPOWER with key functionalities required for UC. These include:

5.5.1 Handling Reserves

Reserves ensure the system can respond to sudden demand changes or generator outages.
MOST can:
• Allocate spinning and non-spinning reserves across committed units
• Ensure reserve targets are met at every time period
• Account for reserve costs in the total objective function
Reserve handling ensures both economic and reliable operation.

5.5.2 Multi-Period Scheduling

MOST schedules generators across multiple time periods:

• Captures minimum up/down times
• Enforces ramp rate limits
• Considers start-up/shut-down costs
• Handles intertemporal constraints

18
By modeling multiple periods simultaneously, MOST produces schedules that are
operationally feasible over the entire day.

5.5.3 Transmission-Constrained OPF

MOST incorporates DC OPF constraints to model the network:

• Ensures that generator dispatch respects line limits
• Accounts for network congestion
• Produces locationally feasible solutions
• Enables study of LMP and congestion costs
This integration is essential for realistic UC studies, as ignoring transmission constraints can
lead to physically infeasible schedules.

5.6 Advantages of Using MATPOWER + MOST

1. Open Source – Free to use and modify, ideal for academic research.
2. Flexible – Users can test new algorithms, constraints, or optimization strategies.
3. Scalable – Capable of handling medium- to large-scale systems using DC OPF
approximation.
4. Transparent – Full access to code and input data allows researchers to understand
every step of the optimization.
5. Educational Value – Provides an excellent learning platform for students studying UC,
SCUC, and power system optimization.

5.7 Summary

MATPOWER and MOST together provide a robust and comprehensive framework for UC
studies:

• MATPOWER handles power flow and OPF, supporting both AC and DC analysis.
• MOST extends MATPOWER to multi-period UC, incorporating technical constraints,
start-up/shut-down logic, reserves, and transmission constraints.
• Input files for buses, branches, and generators allow system definition and scenario
modeling.
• Advanced features of MOST enable realistic day-ahead scheduling, reliability
studies, and market simulations.

This combination allows researchers to simulate complex UC scenarios efficiently without

relying on costly commercial tools, making it a cornerstone of academic research in power
system optimization.

19
 Chapter 6
Mathematical Formulation of Deterministic Unit Commitment

6.1 Decision Variables

In a deterministic Unit Commitment (UC) problem, the first step is defining decision
variables. These variables represent the decisions that the optimization will make regarding
generator operation over the scheduling horizon.

1. Commitment Variables
Each generator has a binary variable 𝑢𝑖,𝑡 representing whether it is online or offline at
time 𝑡:
1 if generator 𝑖 is ON at time 𝑡,
𝑢𝑖,𝑡 = {
0 if generator 𝑖 is OFF at time 𝑡.

These variables determine the on/off status of each generator, forming the backbone of the UC
model.

2. Generation Output Variables

The amount of power produced by each generator is a continuous variable, 𝑃𝑖,𝑡 . Its value
is constrained by the generator’s minimum and maximum operating limits:
𝑃𝑖min ≤ 𝑃𝑖,𝑡 ≤ 𝑃𝑖max

If a generator is offline (𝑢𝑖,𝑡 = 0), its output is zero.

3. Start-up and Shut-down Variables

To accurately account for generator transitions, auxiliary variables 𝑆𝑈𝑖,𝑡 and 𝑆𝐷𝑖,𝑡 are
introduced. They track whether a generator starts up or shuts down in each time
period.
These decision variables collectively define the operation schedule of all generators across
the planning horizon.

6.2 Objective Function Formulation

The main objective in deterministic UC is to minimize the total operating cost while meeting
demand and respecting all technical constraints. The total cost is composed of:
• Fuel costs associated with generator output
• No-load costs incurred while a generator is online but not producing power
• Start-up and shut-down costs
•
Mathematically, the objective function can be expressed as:
𝑇 𝑁

min⁡ ∑ ∑[𝐶fuel (𝑃𝑖,𝑡 ) + 𝐶no-load ⋅ 𝑢𝑖,𝑡 + 𝐶startup ⋅ 𝑆𝑈𝑖,𝑡 + 𝐶shutdown ⋅ 𝑆𝐷𝑖,𝑡 ]

𝑡=1 𝑖=1

By optimizing this function, the UC model seeks a schedule that balances economic efficiency
and system reliability.

20
6.3 Generator Operation Constraints

Generators must operate within physical and operational limits to ensure safe and reliable
power production. These constraints include:

1. Generation Capacity Limits

𝑃𝑖min ⋅ 𝑢𝑖,𝑡 ≤ 𝑃𝑖,𝑡 ≤ 𝑃𝑖max ⋅ 𝑢𝑖,𝑡

This ensures that each generator produces power only when it is online and remains within its
capacity limits.

2. Minimum Up/Down Times

Once a generator is turned on, it must remain online for a minimum duration (𝑈𝑇𝑖 ), and
once turned off, it must remain offline for a minimum down time (𝐷𝑇𝑖 ). This prevents
frequent switching that could damage equipment:
𝑡 𝑡

∑ 𝑆𝑈𝑖,𝑘 ≤ 𝑢𝑖,𝑡 , ∑ 𝑆𝐷𝑖,𝑘 ≤ 1 − 𝑢𝑖,𝑡

𝑘=𝑡−𝑈𝑇𝑖 +1 𝑘=𝑡−𝐷𝑇𝑖 +1

3. Ramp Rate Constraints

up
Generators cannot increase or decrease output instantaneously. Ramp-up (𝑅𝑖 ) and
ramp-down (𝑅𝑖down) limits are enforced as:
up
𝑃𝑖,𝑡 − 𝑃𝑖,𝑡−1 ≤ 𝑅𝑖 , 𝑃𝑖,𝑡−1 − 𝑃𝑖,𝑡 ≤ 𝑅𝑖down

These constraints ensure safe, realistic, and reliable generator operations.

6.4 Network (Transmission) Constraints

Unit Commitment cannot ignore the transmission network, as generator outputs must be
deliverable to load centers without overloading lines. Two main types of constraints are
considered:

1. Power Balance at Each Bus

At every bus, the sum of generator outputs minus load must equal net injections into
the network:
∑ 𝑃𝑖,𝑡 − 𝐷𝑏,𝑡 = ∑ 𝐹ℓ,𝑡
𝑖∈𝐺𝑏 ℓ∈𝐿𝑏

2. Line Flow Limits

Transmission lines have maximum capacities. Using Power Transfer Distribution

Factors (PTDFs), line flows can be expressed as:
−𝐹ℓmax ≤ ∑ 𝑃𝑇𝐷𝐹ℓ,𝑖 ⋅ 𝑃𝑖,𝑡 − 𝐷ℓ,𝑡 ≤ 𝐹ℓmax
𝑖

Including these constraints ensures that generation schedules are physically feasible and
prevents network congestion or overloads.

21
6.5 Reserve and Ramp Constraints

To maintain reliability, reserves must be scheduled:

• Spinning reserves are extra capacity online and ready to respond instantly to
fluctuations:
spinning
∑ (𝑃𝑖max − 𝑃𝑖,𝑡 ) ⋅ 𝑢𝑖,𝑡 ≥ 𝑅𝑡
𝑖

• Ramp reserves ensure that generators can quickly adjust output in response to sudden
changes in demand or contingencies.
These constraints guarantee that the system remains secure under varying conditions,
complementing the economic objective.

6.6 Complete Mathematical Model

Combining all the elements, the deterministic UC problem can be formulated as:
Decision Variables:
𝑢𝑖,𝑡 , 𝑃𝑖,𝑡 , 𝑆𝑈𝑖,𝑡 , 𝑆𝐷𝑖,𝑡

Objective Function:
𝑇 𝑁

min⁡ ∑ ∑[𝐶fuel (𝑃𝑖,𝑡 ) + 𝐶no-load ⋅ 𝑢𝑖,𝑡 + 𝐶startup ⋅ 𝑆𝑈𝑖,𝑡 + 𝐶shutdown ⋅ 𝑆𝐷𝑖,𝑡 ]

𝑡=1 𝑖=1

Subject to:

1. Generator Limits: 𝑃𝑖min 𝑢𝑖,𝑡 ≤ 𝑃𝑖,𝑡 ≤ 𝑃𝑖max 𝑢𝑖,𝑡

2. Minimum Up/Down Times
up
3. Ramp Rate Limits: 𝑃𝑖,𝑡 − 𝑃𝑖,𝑡−1 ≤ 𝑅𝑖 , 𝑃𝑖,𝑡−1 − 𝑃𝑖,𝑡 ≤ 𝑅𝑖down
4. Power Balance at Each Bus
5. Line Flow Limits: −𝐹ℓmax ≤ ∑𝑖 𝑃𝑇𝐷𝐹ℓ,𝑖 𝑃𝑖,𝑡 − 𝐷ℓ,𝑡 ≤ 𝐹ℓmax
spinning
6. Reserve Requirements: ∑𝑖 (𝑃𝑖max − 𝑃𝑖,𝑡 ) ⋅ 𝑢𝑖,𝑡 ≥ 𝑅𝑡

This MILP (Mixed-Integer Linear Programming) formulation balances economic

efficiency, operational feasibility, and reliability, forming the foundation for solving UC
problems using tools like MOST.

22
 Chapter 7: Implementation Using MATPOWER/MOST
7.1 Preparing Input Data
Before solving a Unit Commitment (UC) problem using MATPOWER and MOST, it is
essential to prepare accurate input data. MOST relies on structured files that describe the
power system, generator characteristics, transmission network, and load demands.

7.1.1 Bus Data

The bus data file defines the nodes of the network, representing where generation or load is
connected. Each bus is characterized by:
• Bus Number: Unique identifier for each bus.
• Bus Type: Slack (reference), PV (generator-controlled voltage), or PQ (load bus).
• Load Demand: Real (MW) and reactive (MVar) power.
• Voltage Magnitude Limits: Minimum and maximum voltage levels.
• Shunt Elements: Capacitors or reactors connected to the bus.
This data allows MOST to calculate power injections, voltage profiles, and line flows across
the network.

7.1.2 Generator Data

Generator files provide information about each unit’s capabilities and costs:
• Capacity Limits: Minimum and maximum generation (MW).
• Fuel Costs: Linear or quadratic coefficients representing cost per MWh.
• Ramp Rates: Maximum allowable increase or decrease in output per hour.
• Minimum Up/Down Times: Ensures generator safety and operational feasibility.
• Start-up/Shut-down Costs: Economic impact of turning units on or off.
Proper generator data ensures that MOST produces realistic and feasible schedules while
minimizing total operating costs.

7.1.3 Branch (Transmission Line) Data

Transmission data files describe the electrical network connectivity:

• From and To Buses: Identifies which buses are connected.
• Line Impedance: Resistance and reactance of the line.
• Thermal Limits: Maximum power that the line can carry.
• Transformer Tap Settings: Adjusts voltage levels where needed.
Transmission data is critical for network constraint enforcement and avoiding line overloads
during UC scheduling.

7.1.4 Load Profiles

Load profiles describe system demand over time:

• Hourly or sub-hourly load forecasts.
• Can include realistic variations in demand, such as peak and off-peak periods.

23
 • Renewable generation profiles (wind or solar) can also be included to model variability.
Including accurate load profiles allows the UC model to schedule generators efficiently and
maintain reliability throughout the day.

7.2 Setting Up the MOST Case File

The MOST case file is the central script used to define and solve the UC problem. It combines
all input data and specifies the parameters for the solver.
Steps to set up the case file include:
1. Loading bus, generator, and branch data using MATPOWER format.
2. Defining the time horizon (typically 24 hours).
3. Specifying load profiles and any renewable generation forecasts.
4. Ensuring the system base values (like MVA base) are consistent across all data.
A properly structured case file allows MOST to understand the system topology and
operational requirements, serving as the foundation for the UC optimization.

7.3 Adding UC-Specific Parameters

Unit Commitment requires additional parameters beyond standard OPF:

• Commitment Variables: Binary ON/OFF status for each generator.
• Start-up/Shut-down Costs: Include economic impact of transitions.
• Minimum Up/Down Times: Ensures generators remain on/off for feasible durations.
• Ramp Constraints: Limits on how fast generators can increase or decrease output.
• Reserve Requirements: Specifies spinning and non-spinning reserves to maintain
reliability.
Defining these parameters ensures the MOST model captures the operational behavior of
generators accurately, enabling realistic scheduling.

7.4 Integrating Transmission Constraints

Transmission constraints are integrated using DC power flow approximations:

1. Line Thermal Limits: Ensures that no line exceeds its maximum power flow.
2. Power Transfer Distribution Factors (PTDFs): Maps generator outputs to line flows.
3. Network Congestion Modeling: Captures the impact of overloaded lines on
scheduling decisions.
By including these constraints, MOST can produce physically feasible UC schedules,
preventing generator outputs from violating transmission limits.

7.5 Code Explanation: most_ex6_uc

The most_ex6_uc script is an example MOST implementation. It demonstrates how to define

input data, set UC parameters, and run the optimization. Key components include:
• Data Loading: Bus, generator, branch, and load profile data are loaded into MATLAB
structures.
• Parameter Setup: Generator ramp rates, up/down times, start-up/shut-down costs, and
reserve requirements are defined.
• Solver Call: MOST interfaces with a MILP solver to optimize the UC problem.
• Result Extraction: Outputs include committed units, generation schedules, and system
costs.

24
This script serves as a template for more complex UC studies, allowing users to modify
system size, time horizon, or constraints as needed.

7.6 Solver Configuration (CPLEX/Gurobi/MIPS)

MOST uses MILP solvers to solve the UC problem efficiently. Common solver configurations
include:

• CPLEX or Gurobi: Commercial solvers with fast and robust MILP capabilities.
• MIPS (MATPOWER Interior Point Solver): Built-in solver suitable for smaller
problems.

Configuration steps:

1. Specify the solver in the MOST case file.

2. Set solver options such as tolerance, iteration limits, and solution display.
3. Enable warm-start options if solving multiple periods sequentially for efficiency.

Proper solver configuration ensures that UC optimization is solved quickly and accurately.

7.7 Running the UC Model

After preparing input data, setting up the case file, and configuring the solver, the UC model is
ready to run:

1. Execute the most_ex6_uc script in MATLAB.

2. MOST will formulate the MILP problem including all constraints and the objective
function.
3. The solver determines which generators to commit and their optimal output levels
for each time period.
4. Results include:
o Generator ON/OFF schedules
o Hourly generation outputs
o System total operating cost
o Line flows and reserve allocations

These results allow researchers and operators to analyze economic efficiency, reliability, and
network feasibility of the UC solution.

25
 CHAPTER-8

8.1 CODE

function most_ex6_uc(quiet)
define_constants;
verbose = 1;
mpopt = mpoption('verbose', verbose);
mpopt = mpoption(mpopt, '[Link]', 1);
mpopt = mpoption(mpopt, 'model', 'DC');
mpopt = mpoption(mpopt, '[Link]', 'DEFAULT');
if ~verbose
mpopt = mpoption(mpopt, '[Link]', 0);
end
mpopt = mpoption(mpopt, 'most.price_stage_warn_tol', 1e-5);
if have_feature('cplex')
%mpopt = mpoption(mpopt, '[Link]', 0); %% automatic
%mpopt = mpoption(mpopt, '[Link]', 1); %% primal simplex
mpopt = mpoption(mpopt, '[Link]', 2); %% dual simplex
%mpopt = mpoption(mpopt, '[Link]', 3); %% network simplex
%mpopt = mpoption(mpopt, '[Link]', 4); %% barrier
mpopt = mpoption(mpopt, '[Link]', 0);
mpopt = mpoption(mpopt, '[Link]', 0);
mpopt = mpoption(mpopt, '[Link]', 2);
end
if have_feature('glpk')
mpopt = mpoption(mpopt, '[Link]', 0);
mpopt = mpoption(mpopt, '[Link]', 1e-10);
mpopt = mpoption(mpopt, '[Link]', 1e-10);
end
if have_feature('gurobi')
%mpopt = mpoption(mpopt, '[Link]', -1); %% automatic
%mpopt = mpoption(mpopt, '[Link]', 0); %% primal simplex
mpopt = mpoption(mpopt, '[Link]', 1); %% dual simplex
%mpopt = mpoption(mpopt, '[Link]', 2); %% barrier
mpopt = mpoption(mpopt, '[Link]', 2);
mpopt = mpoption(mpopt, '[Link]', 0);
mpopt = mpoption(mpopt, '[Link]', 0);
end
if have_feature('mosek')
sc = mosek_symbcon;
mpopt = mpoption(mpopt, 'mosek.lp_alg', sc.MSK_OPTIMIZER_DUAL_SIMPLEX);
mpopt = mpoption(mpopt, '[Link].MSK_IPAR_MIO_NODE_OPTIMIZER',
sc.MSK_OPTIMIZER_DUAL_SIMPLEX);
mpopt = mpoption(mpopt, '[Link].MSK_IPAR_MIO_ROOT_OPTIMIZER',
sc.MSK_OPTIMIZER_DUAL_SIMPLEX);
mpopt = mpoption(mpopt, '[Link].MSK_DPAR_MIO_TOL_ABS_RELAX_INT', 1e-9);
mpopt = mpoption(mpopt, '[Link].MSK_DPAR_MIO_TOL_REL_GAP', 0);
mpopt = mpoption(mpopt, '[Link].MSK_DPAR_MIO_TOL_ABS_GAP', 0);
end
if have_feature('intlinprog')
mpopt = mpoption(mpopt, '[Link]', 'dual-simplex');
mpopt = mpoption(mpopt, '[Link]', 'dual-simplex');
mpopt = mpoption(mpopt, '[Link]', 1e-9);
mpopt = mpoption(mpopt, '[Link]', 0);
mpopt = mpoption(mpopt, '[Link]', 0);

26
 mpopt = mpoption(mpopt, '[Link]', 1e-6);
end
casefile = 'ex_case3b';
mpc = loadcase(casefile);
xgd = loadxgendata('ex_xgd_uc', mpc);
[iwind, mpc, xgd] = addwind('ex_wind_uc', mpc, xgd);
profiles = getprofiles('ex_wind_profile_d', iwind);
profiles = getprofiles('ex_load_profile', profiles);
nt = size(profiles(1).values, 1);
mpc_full = mpc;
xgd_full = xgd;
[Link](:, [STARTUP SHUTDOWN]) = 0;
[Link](2) = 1;
[Link](3) = 250;
[Link](3) = 1e-6;
[Link](3) = 1e-6;
mpopt = mpoption(mpopt, 'most.dc_model', 0); % use model with no network
mdi = loadmd(mpc, nt, xgd, [], [], profiles);
mdo = most(mdi, mpopt);
if verbose
ms = most_summary(mdo);
end
mpopt = mpoption(mpopt, 'most.dc_model', 1); % use DC network model (default)
mdo = most(mdi, mpopt);
if verbose
ms = most_summary(mdo);
end
[Link](2, [STARTUP SHUTDOWN]) = [ 200 200];
[Link](3, [STARTUP SHUTDOWN]) = [3000 600];
mdi = loadmd(mpc, nt, xgd, [], [], profiles);
mdo = most(mdi, mpopt);
if verbose
ms = most_summary(mdo);
end
[Link](2) = 3;
mdi = loadmd(mpc, nt, xgd, [], [], profiles);
mdo = most(mdi, mpopt);
if verbose
ms = most_summary(mdo);
end

[Link](3) = 100; % restore ramp reserve

[Link](3) = 10; % constraint and costs
[Link](3) = 10;
mdi = loadmd(mpc, nt, xgd, [], [], profiles);
mdo = most(mdi, mpopt);
if verbose
ms = most_summary(mdo);
end
mpopt = mpoption(mpopt, '[Link]', 1);
[iess, mpc, xgd, sd] = addstorage('ex_storage', mpc, xgd);
mdi = loadmd(mpc, nt, xgd, sd, [], profiles);
mdo = most(mdi, mpopt);
if verbose
ms = most_summary(mdo);
end

27
 Chapter 9: Case Study and Results
Deterministic Unit Commitment Results

9.1 Objective Function

The objective function value obtained from the deterministic Unit Commitment optimization
is:
𝑓 = −4,987,029.64

This represents the minimum total generation cost over the 12-hour scheduling horizon. The
negative value indicates the cost minimization achieved by optimally scheduling generator
commitment and dispatch, while respecting all operational constraints.
Visual: A textual box or annotation summarizing the minimum cost at the top of your
document.

8.2. Generator Commitment Status (GEN_STATUS)

The generator commitment status table indicates which generators were ON (1) or OFF (0)
for each hour:

Generator t=1 t=2 t=3 t=4 t=5 t=6 t=7 t=8 t=9 t=10 t=11 t=12
1 1 1 1 1 1 1 1 1 1 1 1 1
2 1 1 1 1 1 1 1 1 1 1 1 1
3 1 1 1 1 1 1 1 0 0 0 0 0
4 1 1 1 1 1 1 1 1 1 1 1 1
5 1 1 1 1 1 1 1 1 1 1 1 1
6 1 1 1 1 1 1 1 1 1 1 1 1
Observation:
• Generators 1, 2, 4, 5, and 6 are online throughout the 12-hour period.
• Generator 3 shuts down after hour 7, likely due to reduced demand or cost optimization.

28
9.3. Generator Output (PG)

The power output of each generator is shown in MW:

Generator t=1 t=2 t=3 t=4 t=5 t=6 t=7 t=8 t=9 t=10 t=11 t=12
1 186.17 200 181.17 199 190 200 200 200 200 200 200 200
2 65 66.17 65 65 65 70 90 102.5 77.5 77.5 91.5 112.5
3 122.67 122.67 167.67 110 110 110 60 - - - - -
4 -440 -480 -540 -525 -500 -450 -400 -350 -300 -325 -375 -425
5 80 65 60 82 100 70 50 85 100 110 106 95
6 -13.84 26.16 66.16 69 35 - - -37.5 -77.5 -62.5 -22.5 17.5

Observation:
• Generators 1 and 2 provide base and mid-range load respectively.
• Generator 3 is offline after t=7.
• Negative values (Generators 4 and 6) may represent pumped storage or reverse
power flow.

9.4. Ramp Up / Ramp Down Analysis

Ramp values indicate changes in output between consecutive hours.
Observation:

• Generator 3 ramps down to 0 after t=7, reflecting its shutdown.

• Generator 1 operates almost constantly with minimal ramping.
• Ramp up / down patterns highlight flexibility and reserve capability.

29
[Link] Storage System (ESS) SoC

ESS t=1 t=2 t=3 t=4 t=5 t=6 t=7 t=8 t=9 t=10 t=11 t=12
1 196.33 170.17 104 35 - - - 37.5 115 177.5 200 182.5

Observation:

• ESS discharges initially to meet demand, then charges in later hours.

• State-of-charge trends illustrate energy storage utilization for peak shaving and load
balancing.

30
9.6. Locational Marginal Price (LMP)

Bus t=1 t=2 t=3 t=4 t=5 t=6 t=7 t=8 t=9 t=10 t=11 t=12
1 25 30 25 25 25 30 30 30 30 30 30 30
2 25 35 60 39.17 39.17 41.67 30 30 30 30 30 41.67
3 25 40 95 53.33 53.33 53.33 30 30 30 30 30 53.33

Observation:

• Price spikes at t=3 for Bus 2 and 3 indicate congestion or scarcity periods.
• Bus 1 remains relatively stable due to sufficient generation supply.

9.7. Branch Power Flows (PF)

Branch t=1 t=2 t=3 t=4 t=5 t=6 t=7 t=8 t=9 t=10 t=11 t=12
1 16.17 26.17 6.17 24 15 30 60 72.5 59.17 55.83 61.83 72.5
2 235 240 240 240 240 240 230 230 218.33 221.67 229.67 240
3 218.83 213.83 233.83 216 225 210 170 157.5 159.17 165.83 167.83 167.5

Observation:

• Branch 2 experiences congestion (> 230 MW) at several time points.

• Branch 3 is moderately loaded and reduces flow over time.

31
9.8. Lagrange Multipliers (MU_F)

Branch t=1 t=2 t=3 t=4 t=5 t=6 t=7 t=8 t=9 t=10 t=11 t=12
1 - - - - - - - - - - - -
2 - 15 105 42.5 42.5 35 - - - - - 35
3 - - - - - - - - - - - -

Observation:
• Only Branch 2 shows non-zero Lagrange multipliers, indicating active branch flow
constraints.
• These values reflect the marginal cost of relaxing the branch flow limit.

32
 Chapter 10: Conclusion and Future Work
10.1 Summary of Findings
The deterministic Unit Commitment (UC) study performed in this project produced several
meaningful observations regarding system operation, cost behavior, and the influence of
network constraints.

First, the UC model successfully identified the least-cost generation schedule while ensuring
that all operational limits—such as ramping, generator capacities, and minimum up/down
times—were fully respected. The dispatch pattern followed a logical structure: base-load units
operated steadily throughout the day, intermediate units responded to moderate changes in
demand, and peaking units were used only when demand rose sharply. This confirmed that the
optimization model appropriately differentiated between generator types based on cost and
flexibility.

10.2 Importance of Transmission Constraints in UC

Transmission constraints are a crucial element in realistic UC modeling. Their inclusion
ensures that the generation schedule not only minimizes cost but also respects the physical
capabilities of the network.

One significant impact of transmission limits is congestion management. When power cannot
be freely transferred across the system, the optimization must modify generator outputs to
avoid line overloads. These redispatch actions directly affect system costs. Even though they
may increase operating expenses, they maintain grid reliability and prevent potential violations
that could lead to outages.

10.3 Accuracy and Performance of MOST

33
The MATPOWER Optimal Scheduling Tool (MOST) proved to be a highly capable platform
for conducting deterministic UC analysis. MOST efficiently solved the mixed-integer
optimization problem, generating feasible commitment and dispatch schedules across the 24-
hour horizon.
In all cases tested—ranging from unconstrained UC to scenarios that included ramping limits,
startup costs, and network constraints—MOST delivered solutions reliably and within
reasonable computation times. Its internal solvers worked smoothly with integer variables and
handled complex operational constraints without difficulty.
The resulting schedules aligned with expected system behavior, reinforcing that MOST is both
accurate and robust. For research, academic projects, and small-scale practical studies, MOST
offers an excellent environment for analyzing short-term scheduling problems.

10.4 Limitations
While the model performed well, several limitations should be acknowledged.
First, the deterministic formulation assumes perfect foresight of demand and renewable output.
In reality, forecasting errors are unavoidable, especially for weather-dependent sources like
wind and solar. This makes purely deterministic schedules less adaptable to uncertainty.
Second, renewable generation was not explicitly incorporated in the main UC formulation.
Since modern power grids increasingly depend on renewables, excluding them reduces the
realism and applicability of the model.

Another limitation is the absence of security constraints. Without N-1 contingency analysis,
the system is not guaranteed to operate safely if a major component—such as a generator or
transmission line—fails unexpectedly.

Finally, economic market interactions such as bidding strategies, marginal pricing, and market-
clearing mechanisms were outside the project scope. These factors are important in real-world
system operations and can heavily influence optimal dispatch decisions.
Identifying these limitations provides direction for improving and expanding the UC model in
future work.

10.5 Recommended Future Enhancements

To enhance the practicality and robustness of UC models, several extensions are recommended:

34
Stochastic UC: Introducing uncertainty in load and renewable forecasts would allow the model
to generate schedules that are more resilient to real operating conditions. This would support
risk-aware decision-making.

Renewable Integration: Explicitly modeling wind, solar, and other variable resources ensures
that the UC problem reflects the increasing shift toward clean energy sources. This also enables
studies involving curtailment, variability, and forecast errors.

Security-Constrained UC (N-1): Incorporating contingency analysis ensures the generation

schedule remains feasible even if a critical system element fails. This leads to more reliable
system operation and reduces the risk of cascading failures.

Market-Clearing Models: Integrating economic mechanisms—such as pricing, unit bidding,

and congestion costs—would align the UC model with actual electricity market structures. This
is especially useful for evaluating market behavior and economic efficiency.
Implementing these enhancements will make UC studies more comprehensive, adaptable, and
relevant to the evolving needs of modern power systems.

35
 REFERENCES

A. Unit Commitment (UC) & Deterministic UC

[1] A. J. Wood, B. F. Wollenberg, and G. B. Sheblé, Power Generation, Operation, and
Control, 3rd ed. New York, NY, USA: Wiley, 2013.
[2] D. P. Bertsekas, Dynamic Programming and Optimal Control, 4th ed. Belmont, MA, USA:
Athena Scientific, 2017.
[3] C. Wang and S. M. Shahidehpour, “Effects of ramp-rate limits on unit commitment and
economic dispatch,” IEEE Trans. Power Syst., vol. 8, no. 3, pp. 1341–1350, Aug. 1993.
[4] S. Virmani, E. C. Walters, and A. Sheble, “Optimization for security-constrained unit
commitment,” IEEE Trans. Power Syst., vol. 3, no. 2, pp. 421–429, May 1988.
[5] T. Senjyu, Y. Miyagi, A. Yona, N. Urasaki, and T. Funabashi, “A unit commitment problem
by using genetic algorithm based on unit characteristic classification,” IEEE Trans. Power
Syst., vol. 21, no. 2, pp. 1029–1038, May 2006.

B. Transmission-Constrained Unit Commitment (SCUC)

[6] J. A. Momoh, M. E. El-Hawary, and R. Adapa, “A review of selected optimal power flow
literature to 1993. Part II: Newton methods and algorithms,” IEEE Trans. Power Syst., vol. 14,
no. 1, pp. 105–111, Feb. 1999.
[7] F. Bouffard and F. D. Galiana, “Stochastic security for operations planning with significant
wind power generation,” IEEE Trans. Power Syst., vol. 23, no. 2, pp. 306–316, May 2008.
[8] A. Castillo and R. O’Neill, “Computational performance of solution techniques applied to
the security constrained unit commitment problem,” U.S. Federal Energy Regulatory
Commission, Washington D.C., Tech. Rep., 2013.
[9] Z. Li, Q. Guo, H. Sun, J. Wang, and P. Luh, “A new Lagrangian relaxation based
decomposition approach for security-constrained unit commitment,” IEEE Trans. Power Syst.,
vol. 32, no. 3, pp. 2013–2025, May 2017.

C. MATPOWER & MOST

[10] R. D. Zimmerman, C. E. Murillo-Sánchez, and R. J. Thomas, “MATPOWER: Steady-

state operations, planning and analysis tools for power systems research and education,” IEEE
Trans. Power Syst., vol. 26, no. 1, pp. 12–19, Feb. 2011.
[11] C. E. Murillo-Sánchez, R. D. Zimmerman, C. L. Anderson, and R. J. Thomas, “Secure
planning and operations of systems with stochastic sources, energy storage, and active
demand,” IEEE Trans. Smart Grid, vol. 4, no. 1, pp. 222–231, Mar. 2013.
[12] MATPOWER Documentation, “MOST: MATPOWER Optimal Scheduling Tool –
Technical Manual,” PSERC, Ithaca, NY, USA, 2015. [Online]. Available: https:
(MATPOWER website—not included per your instructions)
[13] R. D. Zimmerman and C. E. Murillo-Sánchez, “Matpower User’s Manual, Version 7.1,”
PSERC, 2020.

D. Power System Optimization & OPF

[14] J. Carpentier, “Optimal power flows,” Int. J. Electr. Power Energy Syst., vol. 1, no. 1, pp.
3–15, Jan. 1979.
[15] S. Frank, I. Steponavice, and S. Rebennack, “Optimal power flow: A bibliographic survey
I—Formulations and deterministic methods,” Energy Syst., vol. 3, pp. 221–258, 2012.

36
[16] H. Sun et al., “Review of challenges and research opportunities for optimal power flow,”
IEEE Trans. Power Syst., vol. 33, no. 3, pp. 3410–3425, May 2018.
[17] S. M. Shahidehpour, H. Yamin, and Z. Li, Market Operations in Electric Power Systems.
New York, NY, USA: Wiley, 2002.

E. Mixed-Integer Programming for UC

[18] P. Shah, S. Ahmed, and G. Latorre, “A review of fast algorithms for solving unit
commitment and economic dispatch problems,” IEEE PES General Meeting, pp. 1–8, 2010.
[19] H. Ye, Z. Li, and M. Shahidehpour, “Adequacy assessment of power systems considering
demand response and unit commitment,” IEEE Trans. Power Syst., vol. 30, no. 4, pp. 2235–
2243, Jul. 2015.

37