Consolidating a Varied Generating Fleet

Photo (above): The 288-MW E.B. Campbell station, located on the Saskatchewan River near Nipawin, is one of SaskPower’s projects now managed using Hatch’s Vista DSS.

Learn how Canadian utility SaskPower integrated its hydro and non-hydro generating assets under one management system using Hatch’s Vista Decision Support System.

By Ryan Bannerman, Tryggvi Olason and Richard Allen

Ryan Bannerman is a senior engineer in commercial operations for Northpoint Energy Solutions — SaskPower. Tryggvi Olason is a senior project manager and application developer at Hatch Renewable Power. Richard Allen is a senior water resources and optimization specialist for Hatch Renewable Power.

When utility SaskPower decided to implement a new operations support system to manage all of its generating assets, the western Canadian company had to find a solution that could balance its thermal resources with its hydroelectric fleet. SaskPower worked with Hatch Renewable Power to develop a scheme that allows the utility to optimize both types of facilities.

For the first time in its history, SaskPower now has an hourly optimized dispatch model that includes all types of generation, along with the operating reserves and constraints for dispatch guidance for the operators in its grid control center. The model has particular value in that it provides SaskPower with both short-term real-time operational data and long-term study analysis.

About SaskPower

SaskPower is a vertically integrated utility that serves a large geographic area and has a wide range of generation sources. The utility has a cumulative generating capacity of 3,428 MW, which has historically been dominated by lignite coal-fired sources.

The utility also has significant renewable energy sources, including hydro but also wind, biomass and heat recovery generation, with the world’s first utility-scale “clean coal” carbon capture facility ready to come on line soon.

SaskPower operates seven hydro facilities with a total capacity of about 850 MW in the southern half of Saskatchewan. Inflow is primarily mountain runoff from the South and North Saskatchewan rivers, with releases from the 186-MW Coteau Creek project on Lake Diefenbaker joining the North Saskatchewan River after five to seven days’ travel time. The combined flow then travels through the 255-MW Nipawin and 288-MW E.B. Campbell plants.

Also included on the SaskPower hydroelectric system is the 101-MW Island Falls station on the Churchill River and three smaller plants on the Athabasca system in northwestern Saskatchewan. These Plants are typically dispatched for energy needs only.

Other recent additions have tended toward contracting for additional capacity from independent power producers, whose plants are typically gas-fired co-generation projects. Counting power purchased from other generators, SaskPower’s total available generating capacity is just over 4,280 MW.

With recent increases in demand caused by economic growth in the province of Saskatchewan and the addition of many types of generation to its system, SaskPower needed an operational dispatch planning tool that was not only able to minimize its generation cost to meet load and associated reserves, but also permit the utility to perform ad-hoc studies and evaluations fur future plant upgrades, rehabilitations and system operations.

Understanding the problem

Electricity is an instantaneous product. It must be produced the instant that it is consumed. A control area can neither aid nor abet frequency regulation with neighboring control areas. NERC regulations mandate that interties must be “zeroed out” on a 15-minute basis, meaning that generation equals load. Given that SaskPower has a relatively small control area, unknown variability in both renewable generation and arc-furnace load make dispatch decisions a problem.

During times of high or low runoff, the hydro generation has a significant impact on the dispatch of the thermal generation.

In times of high run-off, quicker response hydro generation is typically base-loaded to wasted energy. This means that regulation must come from other slower, non-hydro responding resources. Variability then makes dispatch a challenge.

Conversely, in times of low run-off, quicker response hydro generation has limited amounts of water for energy generation. They are typically used for generation during the daily peak load during the day, and have the capacity to manage the variability but do not have the energy to back it up. Again variability makes dispatch challenging.

SaskPower’s attempts in the past to develop an operations planning and dispatch model have largely centered on utilizing thermally-oriented models. This left a very weak hydro dispatch and a poor dispatch in general.

Thermal models did not account for reserves on the hydropower facilities, the water travel time between plants, or reservoir elevations and volumes, and, hence, available energy. The models were also unable to dispatch the hydro generation with any of the licensed operating constraints that are unique to hydro, including daily and annual restrictions like environmental concerns for fish spawning, elevation changes after ice recession and water discharge rates.

The province’s geography also poses a challenge in integrating SaskPower’s hydropower assets into an operations support system. Saskatchewan is relatively flat, with a significant travel time between plants. As mentioned earlier, water releases from Lake Diefenbaker on the South Saskatchewan River take more than a week to reach Codette Reservoir downstream.

Therefore, it is important to capture the routing and attenuation of river flow, and short-term hydro scheduling has to consider a two-week horizon.

There are also a number of requirements and restrictions that have to be considered in scheduling hydro generation, as the Saskatchewan River is a multipurpose water resource.

Many new types of generation — each with their own unique set of operating characteristics — made the economic dispatch of the system increasingly difficult.

Finding a solution

Given the makeup of SaskPower’s generating facilities, a thermally-based model of any kind would have been inadequate for the real-time dispatch. Thermal plants are relatively difficult to operate but relatively easy to economically dispatch, whereas hydroelectric plants are relatively easy to operate but relatively difficult to economically dispatch.

SaskPower knew it needed a hydro-based model and had already been using components of Hatch’s Vista Decision Support System (DSS) in its long-term planning, making the decision to extend use of this model to its short-term planning a natural solution.

Vista DSS now allows SaskPower to determine an optimum mid-term plan and hourly resolution hydro-thermal schedule that meets load, reserve and operational constraints. The system uses seven main modules implemented across three databases, which include one for operations called “production,” one for examining new configurations called “tests,” and one for long-term planning called “study.”

Scheduling methodology

Vista DSS is designed to aid dispatchers in the decision-making process of unit generation and water release scheduling. The basic hydro-thermal optimization problem is to resolve the economic tradeoff benefits of current and future power generation, within the operational constraints of the water resource system, considering the inherent uncertainty of future events such as inflows and price.

The unit generation scheduling problem is affected by both short-term (for example, the current week) operational decisions and longer-term (season and annual storage reservoir operations) water use decisions and their economic effects. Therefore, the scheduling problem is decomposed temporally into long-term and short-term scheduling horizons.

The long-term generation scheduler involves primarily reservoir storage management using a coarser time resolution, typically time-steps of a week, with a horizon of one to two years. Short-term scheduling concerns detailed scheduling for a period of several weeks.

Factors that complicate the hydro scheduling problem include:

  • Uncertain inflows and limited capability to store the inflows;
  • Hydraulic-related complications as the distance between linked plants can cause delays; and
  • Non-linear power as a function of head and discharge, or heat rate in the case of thermal plants.

To simplify and manage the problem size, generation resources are usually represented at a plant level in both the long- and short-term scheduling problem. Or, with the unit dispatch commitment and loading, treated as a sub-problem. Certain characteristics are very challenging and require special strategies, including unit commitments, plant retirements, rough zones and gaps, units that must be operated at a singular loading, minimum run times, start-up and shut-down costs, conditional constraints and reserve requirements.

The Vista DSS system includes detailed models of all of SaskPower’s hydroelectric generation facilities and their associated river systems, as well as constraints imposed on their operations. Complicated hydraulic connectivity can be represented, including storage reservoirs with multiple outlets and plants that receive water from multiple sources.

Integrating the system

Hydroelectric plants, spillways and flow release structures can be represented in detail, including power as a function of head and discharge, maximum and minimum limits as a function of head, rough zones, and eligibility of various reserve duties. Tailwater elevations can also be modeled as a function of total discharge and/or downstream water levels if there is a backwater effect. Various spill structures can be assigned to spill flow arcs, including tainter gates, orifices, stoplog structures and valves. Uncontrolled diversion tunnels and channels are also supported.

The thermal generation at SaskPower was modeled on a separate system within Vista DSS. This enabled the Vista DSS scheduling engine to minimize the cost for meeting load and reserve requirements by considering all of the dispatch constraints for the utility’s system at once.

Each of the thermal plants have a unique fuel with a fuel cost, measured in dollars-per-gigajoule, applied to the fuel volume. The model then minimizes this cost, while meeting the demand against all of SaskPower’s other resources, under all of the unit, plant and system dispatch constraints related to the utility’s grid system.

Any individual thermal heat-rate curve, relating to megajoules per megawatt hour versus megawatt is converted into a “thermal hill curve” of megawatts versus gigajoules. The model then uses this curve to economically dispatch the generator output.

Gas-fired and co-generation facilities have been modeled by having the “upstream” plant contain the gas turbines and the “downstream” plant containing the heat recovery steam generator (HRSG) or once-through steam generation (OTSG). All of the flow from the upstream gas turbines flows through the downstream steam plant. The fuel cost is applied to the gas turbines, which accounts for the plant input fuel cost, while an output thermal hill curve for the HRSG can be constructed based on all of the flow through the gas turbines. The output from the steam plant is essentially free.

Minimum run time, minimum stop time, maximum stop time, start cost, shutdown cost and start fuel for each unit are also modeled, allowing for flexibility in the modeling of the thermal generation and use of the model to determine the most economic output from each generation source.

External generation resources — including independent power producers and future sources — must also be considered and are represented within the system as power-purchase opportunities if they have a dispatchable component.

Forecasting and management

Accurate load forecasts are an integral part of scheduling generation resources at SaskPower. The system allows the utility to define multiple customers, with each customer mapped to a load bus. Long- and short-term customer load forecasts can be generated or imported for each customer, with generation then optimally dispatched to meet all system load, reserve and constraints.

Reserves are defined to enable the production of power on short notice, since supply must equal demand at all times. For that reason, the SaskPower control area is obligated to carry reserves to ensure adequate operations and maintain reliability. Reserves are divided into four major types:

  • Operating reserve, which is the reserve necessary to cover the unexpected loss of the largest unit of the grid;
  • Spinning reserve, which makes up a minimum of 40% of the operating reserve. It is typically made up from unutilized on-line capacity that is available within 15 minutes;
  • Non-spinning reserve, which typically comes from off-line, quick-start generating units available within 15 minutes; and
  • Regulating reserve, which is the capacity maintained to cover short-term change in internal load and to regulate tie-lines with neighboring utilities.

SaskPower maintains what is known as symmetric regulation, meaning that the capacity assigned for non-spinning reserve (also known as automatic generation control, or AGC) must be capable of increasing or decreasing at a given time by a similar amount.

Each unit model component includes “reserve availability,” which indicates whether it can meet each of the four reserve types, and also a limiting amount of each reserve type. For example, units with slower ramp rates can be limited to a defined megawatt value rather than the difference between maximum and minimum capacity.

As the system optimizes energy dispatch, it also optimizes reserve dispatch. If reserves are inadequate as a result of the energy dispatch, the model will bring on generation in an economic fashion to meet both the energy and reserve requirements.

Planning and scheduling is a shared responsibility between SaskPower and Northpoint Energy Solutions (NEC), which is responsible for water management, marketing and the Generation Control Centre (GCC). The GCC is responsible for the dispatch of all units for the next few days and for providing the dispatcher with detailed schedules and contingency plans.

Currently, Vista DSS data is reviewed by SaskPower personnel on a weekly basis versus the current spreadsheet-based methodology to allow the system operators some comfort with this new approach to dispatching generation. Constraints and adjustments are expected along the way.

NEC performs an annual long-term analysis and coordinates water management with the Saskatchewan Water Security Agency to establish targets for water releases from seasonal reservoirs. Once per week, or as needed, long-term water management decisions are reviewed and adjustments are made to constraints. These in turn are key drivers for the short-term generation schedule.

Long-term water management decisions concern primarily the use of Lake Diefenbaker’s seasonal storage based on Vista DSS analysis, with recommendations on flow release and the value of water in storage for the current week and expected water level trajectory for the rest of the year. The lake is the utility’s upstream seasonal water storage reservoir and influences the total amount of water that flows through the other reservoirs at any given time, making it of particular importance to SaskPower’s hydro operations.

Meanwhile, short-term scheduling is executed in both manual and automatic modes. In automatic mode, the time horizon moves with the clock every hour, loading data and forecasts. The model then utilizes this as a starting point for a re-optimization, and results are published. The process is repeated at the beginning of every hour. GCC then uses the results as a guideline for the future hour’s dispatch.


Implementation of Hatch’s Vista DSS has allowed SaskPower to manage its generating facilities with improved water management and better cycling of units and reserves, yielding significant economic benefits.

In addition, the ability to optimize schedules with maintenance outages and to respond to forced outages is important. The system provides a centralized location for all hydraulic data and automatic computation of power discharges, local inflow and inflow forecasts, giving SaskPower the ability to consolidate management of all its resources for the first time.

More HR Current Issue Articles
More HR Archives Issue Articles
Previous articleModeling Potential Turbine Upgrades for the Brazeau Plant
Next articlePakistan’s Winds Blow Slow

No posts to display