Power Optimisation Ltd

POWEROP: UNIT COMMITMENT SOFTWARE FOR POWER SYSTEM SCHEDULING AND ECONOMIC DISPATCH

INTRODUCTION

Electricity generating companies and power system operators use unit commitment software for power system scheduling, to determine how best to meet the changing demand for electricity, which has daily and weekly cycles as well as longer-term variations. Their short-term optimization problem is how to schedule generation to minimise the total fuel cost, or to maximise the total profit, over a study period of typically a day, subject to a large number of operational constraints.

In the longer term, an electricity company must also purchase fuel and trade fuel and power contracts, while satisfying any emission constraints that may apply.

If an electricity company is responsible for meeting the demand for electricity, its most important short-term constraint is that the total generation must equal the forecast demand in each time interval (typically 30 or 60 minutes long). Alternatively, a generating company in a deregulated electricity market may not need to match demand exactly, but must decide at what prices to offer its generating plant to the market at different times of the day.

There are two closely related short-term optimization problems, unit commitment and economic dispatch. Unit commitment determines when and which generating units should be started up or shut down. Economic dispatch determines the power output of each scheduled generating unit at each time-point.

Unit commitment is a very challenging optimization problem because of the enormous number of possible combinations of the on and off states of all the generating units in the power system over the study period. Similar optimization problems must also be solved over longer study periods, such as a year, for planning purposes.

Power Optimisation is a UK company that develops Powerop, a unit commitment and economic dispatch optimization program developed through many years of practical use by the electricity industry. Powerop simultaneously determines both unit commitment and economic dispatch decisions, which improves the quality of the resulting generation schedules.

Powerop includes an option called Planner for longer-term planning studies. Planner automates repeated short-term runs of Powerop to cover longer study periods, allocating limited resources such as fuel supplies and emission allowances across the whole study period while taking account of scheduled maintenance and random breakdowns of generating units. The short-term runs are linked by the allocation of these limited resources, ensuring that long-term constraints are respected across the entire study period.

Powerop has been used operationally for self-scheduling by electricity generating companies and as the optimization engine for scheduling and dispatching generating units by power system operators. Typical applications for system operators include Reliability Unit Commitment (RUC). Powerop is currently used operationally for RUC by the system operator of a large interconnected power system in North America. This application is used to identify generating units that should be started in advance when forecast conditions indicate a risk that demand and reserve requirements may not otherwise be met reliably.

Powerop’s optimization simultaneously determines generator commitment, dispatch levels and reserve provision across multiple time-periods, with commitment, energy and reserves co-optimized while respecting inter-temporal constraints such as ramp rates and minimum up- and down-times. These decisions are determined within a single integrated optimization problem, and multiple reserve products with different response times can be co-optimized within the same optimization.

Powerop is suitable for optimization studies and operational scheduling involving power systems with many generating units and complex operating constraints, requiring large-scale mixed-integer optimization. Its proprietary multi-phase MILP algorithm allows such problems to be solved within practical run times for operational use. The algorithm is designed to scale efficiently as the number of generating units and system constraints increases.

In practical studies, Powerop’s multi-phase option produces high-quality schedules in a few minutes on a laptop computer for real unit commitment problems for a weakly interconnected power system with about 70 generating units, 60 half-hourly intervals, and several overlapping zonal transmission constraints, while co-optimising energy and multiple reserve categories. With the standard MILP option, similar-quality schedules for such problems take up to 5 times longer.

Powerop also contains features that are particularly useful for the operational challenges faced by weakly interconnected or isolated power systems.

ADVANTAGES

Powerop uses a proprietary optimization algorithm developed by Power Optimisation, based on a multi-phase version of the Mixed Integer Linear Programming (MILP) method. This has the following advantages:

• The generation schedule produced by Powerop is always feasible, that is it satisfies all the defined constraints provided the data are self-consistent. If the data or constraints are not self-consistent and a feasible solution does not exist, the lower priority constraints are relaxed automatically so that a usable schedule is still produced. This is particularly important for operational use, where imperfect forecasts or data inconsistencies can occasionally make the optimization problem (as specified in the input data) infeasible.
• The generation schedules produced are typically significantly more economical than those obtained using manual scheduling or simple priority-order methods, leading to substantial reductions in operating costs.
• A wide variety of constraints and plant types can be modelled, including complicated scheduling constraints, non-linear cost curves, energy-limited plant, power-purchase agreements and emission constraints.
• It is relatively easy and quick to introduce new constraints and features and to modify Powerop as circumstances change.
• Powerop is robust to changes in operating conditions and relative fuel costs, because no prior assumptions are made about the nature of the solution.
• As a result of using a proprietary multi-phase version of the MILP method, Powerop finds very good feasible schedules significantly faster than a conventional single-phase MILP formulation. It also has an option to use the standard MILP method for comparison purposes.
• For long-term planning purposes, the use of Powerop's Planner option means that long-term optimization is combined with a very accurate model of short-term generator scheduling. This makes its results much more realistic than those from the approximate models that are often used for such purposes. This also means that there is no mismatch between short-term generator schedules and longer-term plans.

FEATURES

The objective of the optimization is to minimise the total fuel costs over the study period whilst satisfying the appropriate constraints. Alternatively, the objective of the optimization can be to maximise 'profit', defined as the total revenue from electricity sales minus fuel costs. The study period is divided into time intervals chosen by the user, which are usually half-hourly or hourly (but longer or shorter time intervals can also be used). For most power systems, study periods of up to a week with half-hourly time intervals can be solved to good accuracy by Powerop within an acceptable run time. For very large power systems and / or longer study periods, Powerop's Planner option can be used to automatically perform repeated linked short-term Powerop studies which can optimize limited resources over the long-term.

Powerop takes into account any combination of the following features and constraints of the power system being modelled:

Types of Generating Units

Powerop can model the operation of thermal, hydro-electric and renewable generating units and electricity contracts with other companies. Thermal generating units can include coal-fired, oil-fired, gas-fired and nuclear-powered steam turbines, and combustion turbines burning distillate (which are often called gas turbines). Thermal units can also include dual-fired generating units, which can use one of two alternative fuel types. Powerop can model and optimize the use of mixtures of different fuels and/or gas contracts in the same generating unit or group of units. The hydro-electric units can include conventional hydro units and Pumped Storage units. Gas-fired units and hydro-electric units are treated as energy-limited plant. Energy storage units, such as Battery Energy Storage Systems (BESS) and pumped storage, can be modelled and their operation optimized. Upper or lower energy limits may also be applied to other types of generating units and may be used, for example, to represent limited fuel supplies or 'take or pay contracts'. Renewable generation, such as wind farms and solar power, can be modelled as pseudo-generating units with zero fuel costs and time-varying availabilities, which might represent the forecast total wind power or solar power in a particular region.

System Constraints

• The total output of all the generating units must be equal to the forecast value of the system demand at each time-point. The penalty for not doing so is set by the user, so this constraint may be made 'hard' or 'soft', depending on the user's requirements.
• The total spinning-reserve from all the generating units must be greater than or equal to the spinning-reserve requirement of the system. This can be a fixed requirement in MegaWatts (MW), a specified percentage of the largest on-load output of any generating unit, or a combination of the two. Again, the user can choose how strongly to enforce this constraint. The purpose of the spinning-reserve requirement is to ensure that there is enough spare capacity from the generating units that are on-load or 'spinning' at any time to cover the accidental loss of any individual generating unit, or to satisfy demands that in practice are higher than their forecast values. The precise definition of how much spinning reserve a particular generating unit supplies can be customised to the user's requirements. Several different spinning reserve requirements may be co-optimized, with each reserve requirement being over a different time-scale.
• Powerop can also model standing reserve. This is reserve that is provided by generating units that are currently off-load but which can be started-up and dispatched within a relatively short time, such as 10 minutes.
• Powerop can also impose a downward-reserve requirement. This ensures that there are sufficient generating units running above their Minimum Stable Generation levels at all times to allow the total output to be quickly reduced by a specified number of MW. (There may be a variety of reasons for such a requirement - for example to cover the possibility of demands in practice being lower than their forecast values.)
• Powerop can equalise the outputs of identical generating units at any one time, but only if this does not cause violations of any other constraints.

Transmission Constraints

• The user can define groups of generating units, which are in export- or import-limited transmission-constrained zones. The net power injection from each zone (total generation minus the forecast local demand) must not exceed a specified transfer limit. Reserve can also be included in the definition of export transmission constraints.
• Transmission constraints may alternatively be represented using linear sensitivity factors, allowing flows on transmission circuits or interfaces to be expressed as linear combinations of the net injections (generation minus demand) at different locations in the power system. This formulation is equivalent to the use of Power Transfer Distribution Factors (PTDFs) in DC load-flow models.

Cost Characteristics of Generating Units

• Each generating unit has a 'no-load' or fixed operating cost and a number of incremental operating costs, which can define a non-linear profile of operating costs. These costs can alternatively be expressed as fixed and incremental heat rates multiplied by a fuel cost, in which case the fuel costs can vary over the study period. The incremental costs or heat rates modelled by Powerop do not have to be monotonically increasing with increasing power output.
• Each generating unit has either a single start-up cost, or a number of warmth-dependent start-up costs, corresponding to a number of warmth states (for example hot, warm and cold) of each generating unit, determined by the time that particular unit has previously been off-load. Powerop is able to take these into account in its optimization, even if start-up costs do not increase steadily with warmth state.
• Powerop is also able to model works power, which is the power taken from the electricity grid when running a particular generating unit or before a generating unit is synchronised with the grid. Like start-up costs, works power can be warmth-dependent, depending on the warmth condition of the generating unit when it starts-up, as determined by the time that unit has previously been off-load.

Scheduling Constraints on Individual Generating Units

• Each generating unit can have minimum on and off times.
• There can be upper limits on the numbers of start-up events of each generating unit per day and also over longer periods.
• The user can specify inflexible running of generating units. This forces a generating unit to run over a specified time-period, with an output not less than a specified value. Alternatively, the output of a generating unit can be forced to be equal to a specified value over a specified time-period.

Dispatching Constraints on Individual Generating Units

• When a generating unit is on, its power output must be at or above its Minimum Stable Generation value, except when the generating unit is starting up or shutting down.
• The output of a generating unit must not exceed a specified maximum value.
• In Powerop's input data, the user can specify capacity restrictions on generating units. These reduce the maximum outputs of the units over specified time-periods. Alternatively, the user can specify different values of the available capacities of generating units in different time-periods ('profiling the availabilities').
• Each generating unit has multiple-segment spinning-reserve characteristics, which depend on the output of that unit, and also on the category of reserve being modelled. Powerop allows the user to specify requirements for several reserve categories in the same study.
• Generating units can be subject to warmth-dependent run-up rates, which depend on the time that particular unit has previously been off-load. For example, different run-up rates can be specified to Powerop for the 'hot', 'warm' and 'cold' states of a generating unit. When such a unit starts-up, it begins operating at a power output at or below its Minimum Stable Generation (MSG), and it runs-up to MSG by following a non-linear run-up profile, which depends on its warmth state at start-up. Above MSG, this profile also defines an upper limit to the power output. In a similar way, run-down rates (which are not warmth-dependent) apply when a unit shuts-down.
• Generating units are also subject to maximum loading and deloading rates, which apply when running continuously above Minimum Stable Generation.

Station Constraints

• There can be 'station synchronising intervals' and 'station desynchronising intervals' at some power stations. These are minimum time gaps between the start-ups and shut-downs of generating units at the same power station.

Energy Storage Units and BESS

• Powerop can optimize the use of energy storage units, including Battery Energy Storage Systems (BESS) and Pumped Storage hydro-electric units, taking account of the energy lost in their storage and generating cycle. In other words, Powerop optimizes the times and amounts by which to store energy and/or generate from the energy storage units.
• For BESS, Powerop can also be used to recommend the optimal levels for each of the Reserve and Frequency Response Services that BESS can provide. In the British electricity market, Reserve Services include both Positive and Negative Balancing Reserve (PBR and NBR) and Quick Reserve (PQR and NQR). Frequency Response services include both High and Low Dynamic Containment (DCH and DCL), Dynamic Moderation (DMH and DML) and Dynamic Regulation (DRH and DRL). Services can be stacked in accordance with the National Energy System Operator's (NESO's) rules on service delivery and the capabilities of the BESS. The recommended Reserve and Response levels can be bid into the day-ahead procurement auctions via the NESO's Single Markets Platform. After the auction results have been declared and the contracted Reserve and Response levels are known, Powerop can be used to regularly re-optimize the charging and discharging profile of the BESS (with the objective of maximising revenues) as wholesale power prices evolve, both day-ahead and within-day. At all times, Powerop ensures that the BESS stored energy level remains compliant with NESO's state of energy management rules given the contracted Reserve and Response levels and the recommended power exports and imports.
• Click here to view a brochure on using Powerop to optimize BESS.

• In the case of Pumped Storage units, the water stored in the upper reservoir of a Pumped Storage station can be kept between specified upper and lower limits, to prevent spillage or drainage of the reservoir, allowing for any inflow of water into the reservoir. The user can specify to Powerop the minimum, maximum and target values for the reservoir level at the end of the study period, and possibly also target levels at various times within the study period. The user can also determine how firmly those target levels are enforced by choosing appropriate penalty costs.

Electricity Contracts

An electricity contract with another company can be modelled in Powerop as a pseudo-generating unit or as a 'demand' unit. This option offers all the features described above, for example ramp rates, energy limits, and minimum on and off times, which can be useful if the electricity contract contains such features.

Powerop also has an option for the direct modelling of the electricity contracts which are offered on power exchanges or by electricity brokers. Electricity contracts (also known as power contracts) may be of type 'buy' or 'sell'. Electricity contracts may optionally be grouped, so that if one contract in a group is accepted then all the other contracts in that group must be accepted. This direct modelling of electricity contracts allows Powerop to consider a much larger number of such contracts than if they were modelled as pseudo-generating units. This provides guidance to the electricity company as to which contracts to accept and at what volumes, in order to maximise profits, whilst taking into account the knock-on effects of accepting those contracts on the outputs of the physical generating units, including the effects of ramp rates.

Initial Conditions

• At the user's option, Powerop can either use specified values of the initial conditions or it can choose its own initial conditions. Using specified values of the initial conditions would be appropriate for short-term operational use. The option of letting Powerop choose its own initial conditions is suitable for longer-term planning purposes.

Import and Export across an Interconnector

• Powerop can be used to optimize the import and export of power from and to neighbouring electricity utilities via one or more interconnectors. Powerop can also be used in a 'what-if' mode, to evaluate the costs or benefits of proposed power transfers across the interconnector.
• Alternatively, generation and demand may be modelled on both sides of an interconnector, with user-specified half-hourly or hourly values for the limits on the flow across the interconnector, which then behaves like a transmission constraint. It is also possible to treat the demand requirements as being 'harder' on one side of the interconnector than the other.

Modes of Operation

Powerop is able to use pseudo-generating units to model different modes of operation of generating plant, with logical constraints defined by the user preventing simultaneous operation of any incompatible modes. For example, the user might define different pseudo-generating units in Powerop's input data for the following modes of operation of a Combined Cycle Gas Turbine (CCGT) power station that has two gas turbines and one steam turbine: Mode A is Gas Turbine 1 operating in 'open-cycle mode', Mode B is Gas Turbine 2 operating in 'open-cycle mode', Mode C is Gas Turbine 1 operating in 'combined-cycle mode' with the Steam Turbine, Mode D is Gas Turbine 2 operating in 'combined-cycle mode' with the Steam Turbine, and Mode E is Gas Turbines 1 and 2 operating together in 'combined-cycle mode' with the Steam Turbine. The thermal efficiencies, ramp rates and other data items of these modes will not all be the same. This can be modelled by Powerop using pseudo-generating units and appropriate logical constraints specified by the user.

Longer-Term Issues

When performing a planning study over a period of, say, one year, some considerations apply over all or most of the year, rather than at one particular time. Such issues include:

• Emission limits from various pollutants in fuels.
• Fuel-supply contracts (particularly for gas) and fuel constraints.
• Electricity contracts.
• Maximum numbers of start-ups for individual generating units (particularly for gas turbine plant, which commonly have a requirement for a maintenance overhaul after a certain number of start-ups).
• Scheduled overhauls and random breakdowns of generating units.
• Powerop's Planner option allows for scheduled overhauls by means of its input data and it can model random breakdowns probabilistically, including an optional variance-reduction capability.
• The above issues are dealt with by Powerop's Planner option optimizing the limited resources over the whole long-term study period, and then allocating them to subsequent shorter-term Powerop studies. Planner then collates the results of those Powerop studies together for user-friendly presentation to the user. All of the previously-described tasks are performed automatically by Planner. This means that long-term optimization is combined with a very accurate model of short-term generator scheduling. This makes Planner's results much more realistic than those from the approximate models that are often used for long-term planning purposes. This also means that there is no mismatch between short-term generator schedules and longer-term plans.

USE BY POWER SYSTEM OPERATORS

Many of the features described above make Powerop suitable for use by power system operators to schedule and dispatch their power systems in a secure and economic way, including applications commonly described as Reliability Unit Commitment (RUC) and Residual Unit Commitment. Relevant features include:

• Multi-period unit commitment with minimum up/down times, start-up categories and other inter-temporal operating constraints such as ramp rates.
• Reserve requirements, including co-optimization of multiple reserve constraints over different time-scales.
• Zonal and interface constraints using user-defined groups of generation and demand, with import limits, export limits or equality constraints. These may include reserve in the group definition where required.
• Faster run times than the standard MILP method for large power systems, due to Powerop's proprietary multi-phase MILP algorithm for rapidly finding good feasible schedules.

Powerop evolved from earlier unit commitment and dispatch software developed by Power Optimisation. This software was used in operation for the unit commitment and dispatch of the Northern Ireland power system for over a decade prior to the introduction of the Single Electricity Market (SEM) on the island of Ireland. Powerop subsequently evolved to support production cost modelling and planning studies, and for use by major generating companies for self-scheduling in the British electricity market under the British Electricity Trading and Transmission Arrangements (BETTA).

Operational use: Powerop is currently used by the system operator of a large interconnected power system in North America for Reliability Unit Commitment. This application involves determining which generating units should be started in advance when generator self-schedules indicate that there is a significant risk of not being able to satisfy reliably the forecast demands and reserve requirements.

USE UNDER BETTA

Powerop has some special options for use by generating companies that are self-scheduling under the British Electricity Trading and Transmission Arrangements (BETTA) in the British electricity market.

Companies using Powerop under BETTA schedule their generating units against half-hourly 'Net Contract Positions' in MegaWatt Hours (MWh). Powerop can model general market prices for electricity, and can also be used to model individual 'buy' and 'sell' contracts for electricity with their particular characteristics and prices (see the section on electricity contracts above). This provides guidance to the electricity company as to which contracts to accept and at what volumes, in order to maximise profits, whilst taking into account the knock-on effects of accepting those contracts on the outputs of the physical generating units, including the effects of ramp rates.

Powerop has an option to calculate the minute-by-minute output profiles for each generating unit for the 'physical notifications' that are required under the BETTA rules. These output profiles satisfy the constraints on the generating units, whilst minimising imbalances between the total integrated output profiles and the Net Contract Positions. The Powerop study period can be up to several days long, with half-hourly time intervals. Longer-term studies are possible using the Planner option.

USER INTERFACE AND INTEGRATION WITH OTHER SOFTWARE

Powerop runs on a computer with a Windows operating system and is supplied with a sophisticated user-interface based on Microsoft Excel workbooks. Alternatively, Powerop can be used as a solution engine communicating via simple text files with a user-interface developed by the customer. This means that it is straightforward to integrate Powerop with the customer's other software systems. Powerop's optional Excel user-interface is recommended for an initial evaluation of Powerop, because it is easy to use and understand.