Energy portfolio managers and trading groups who strive to operate efficiently and profitably find the uncertainties inherent in the power market very challenging. For GENCOs that operate within an RTO, where day-ahead and real-time markets are collectively optimized, the complexities are no less daunting than for a traditional, load-based utility.
If anything, the need to model, analyze, and plan intensifies due to the competitive nature of the RTO trading environment, especially when decisions impact multi-million-dollar fuel and maintenance budgets.
In this article, we introduce seven broad types of analysis and modeling needed to make smart, impactful decisions in the context of an ISO market. These are not new, but they may become neglected or ignored as the typical ISO DA/RT workflow “goes on autopilot.” Therefore, to continue maximizing asset values, be sure to keep these analytical disciplines on the front burner.
Obviously, there is much ground to cover here, so future associated blog posts will further unpack and explore these areas in greater detail.
1. Fuel Management & Forecasting
Fuel planning, forecasting, and optimization are central to good portfolio management. Generation modeling and analytics are needed to determine how much gas to nominate a day ahead or to lock in bal-month given fuel and power price forecasts. Models can (and should) help answer complex questions like:
- Which pipeline(s) should be scheduled, given uncertain weather forecasts;
- If there are potential Operational Flow Orders (OFOs), what balancing penalties might we incur?
- How might gas scarcity cause LMP spikes (which affect all generators)?
- Given multiple fuels and fuel-switching capabilities, how can transportation costs be optimized to produce more cost-efficient and accurate burn forecasts?
- What fuel or coal blend is optimal under various LMP forecasts due to changes in heat rate and high limits of various blends?
- How are gas units affected by the differences between gas and power day? This discontinuity can cause GENCOs to over or under-nominate day-ahead gas to meet a generation plan based on LMP forecast prices, leaving them exposed to spot gas and imbalance penalties.
- How should we optimally dispatch a generator tied to an LNG terminal to minimize transportation costs, given the ship schedules and tank volumes? There’s a similar problem for coal units with a given shipping schedule and pile limits.
Scenario planning and timely reporting on fuel burn and costs are also critical to help plan more effectively and maximize revenue.
Co-optimization of fuel and power networks is complex. Unit Commitment and Economic Dispatch (UCED) modeling broaden your field of view to help you decide how much fuel to procure or stage before it becomes urgent and costly.
2. Maintenance Optimization
Market participants typically have some form of LMP forecasting capability. One of the many benefits of accurate price forecasting is to determine the best time to take an outage for low-cost units. Simulating an out-of-service unit and comparing profit results with model runs that keep that unit online provide a quantifiable basis for outage scheduling. Similar analysis can help determine, for example, whether to nurse a tube leak a little longer or fix it immediately based on expected LMP price increases.
A longer-term UCED model can help determine the value of Long-Term Service Agreements (LTSA) that typically limit annual run-time and start-ups by imposing penalties when limits are exceeded.
Monte Carlo simulation of a combined cycle generator, using various start and VOM costs, can assess whether the cost of an LTSA is economical. Alternatively, if an LTSA already exists, a good model can help determine what values to use for start-up and VOM costs so that utilization is kept within contractual limits.
3. Operation & Bidding of Complex Assets
PCI has written about storage optimization for batteries and pumped storage hydro. Models like GenTrader can predict when to charge and discharge these resources and define when ancillary services are most profitable under various price scenarios.
However, these are not the only complex assets that benefit from running an economic model for bidding and operations. Co-optimization of distributed energy and demand response programs, multi-stage generation commitment, and committing large units with high minimum load and long cycle limits also demand more in-depth analysis.
ISO’s typically look ahead one day, but planning the commitment for some plants is a multi-day exercise. Many market participants prefer to decide whether to keep a combined cycle or coal unit running over a long weekend if prices are expected to support it rather than allowing the ISO to decide.
4. Environmental Compliance
If your firm operates in a region affected by the Cross-State Air Pollution Rule (CSAPR), then you need to consider the cost of NOx and SOx credits, which each have annual allowances. NOx credits also have associated seasonal allowances.
At the end of each year or ozone season, the amount of allowances for a plant must be at least equal to its emissions for that period. Example:
- A source that emits 10,000 tons of SO2 annually must hold at least 10,000 allowances that can be used in that year.
- As the year progresses, it is important to update the model and adjust incremental bidding costs appropriate to the remaining credits and their replacement costs. UCED models that include emissions costs are crucial in these situations.
5. Portfolio Resource Mix
This often falls under the control of long-range planners who use models to look decades ahead at load growth and duration curves, policy trends, and fuel availability to figure out what needs to be added to the portfolio and when. However, there are often shorter-fused decisions around portfolio planning in an ISO that require UCED modeling, particularly when ISO rules change.
For example, the recent market change to 5-minute settlements triggered several companies to analyze the value of faster ramping peakers. Likewise, the advent of utility-scale batteries has many shops running the numbers on combined solar-battery installations in the near term.
6. Post Analysis
There are as many definitions of Post Analysis (PA) as there are market participants. For some, PA means determining the fuel cost of physical sales to allocate back to ratepayers. Others may be more concerned with how optimally the load was served in their control area. Many ISO participants run a model to assess whether their DA awards were fairly allocated given market clearing prices.
7. Trading & Position Management
Market participants do more than just bid load and offer generation to the DA market. Bilateral deals can help reduce power price exposure, but they can also hurt the bottom line if generation costs are not properly estimated. Whether it’s a bal-week on-peak or a 24/7 prompt month product, effective models can help you determine the correct price to sell power at the right point in the stack, helping to ensure the most reasonable margin.
Integrating a generation model with an Energy Trading and Risk Management (ETRM) system provides distinct advantages in automating position management, and that integration doesn’t have to be complex. Successful GENCOs now consider portfolio optimization as a core component of their ETRM.
No matter how you cut it, running a Post Analysis model is a best practice that will help you uncover problems and improve both financial and operational efficiency.
We have only scratched the surface here by highlighting some scenarios and uses for these types of analyses. Future posts will focus on the how and why in each of the seven categories we’ve touched on and will include helpful examples, tips, and more details on some of the tools used in the modeling field.