by Planning Engineer
Some of the Climate Etc. denizens have requested a post on the generation planning process to help them better understand cost issues surrounding the large scale addition and integration of renewable resources.
The major takeaway is that differing types of generating resources bring diverse sets of costs and benefits to the power system so that they cannot be compared solely based on a cost per Megawatt (MW) produced basis. As this post will conclude, it matters very much when energy is generated, where the energy comes from and how well it works to support the system. For large scale bulk projects, the average cost of solar and wind will need to be significantly below the average cost of conventional generation well before wind and solar can begin to approach general competitiveness on a cost basis.
Generation Planning 101
Traditional Generation Planning starts with a Load Duration Curve (LDC) as illustrated in the accompanying figure. The plot is made by sorting hourly load/demand levels from highest to lowest as you move across the graph from left to right. In the graph below the peak demand is just below 70,000 MWs and the minimum is above 20,000 MWs. It cannot be seen from the chart, but the highest values will occur during the peak daily hours concurrent with extreme temperatures. The lowest values usually occur at night during mild weather. Values from peak hours with mild weather and night hours with more extreme weather get mixed in the middle. The curve below is fairly typical for most large power systems. Energy is the most valuable during hours on the left side of the curve and the incremental costs generally decreases with lower load levels.
A generation planner needs to make sure that there is enough generation capability on line to meet thepeak demand and provide for reserve margins under forecast growth conditions. The limited peak hours account for a significantly disproportionate share of power system costs. The peak demand hour(s) on the left side of the curve drive the need for new generation. Once the need level is established, generation is selected to work with the existing system resources to provide for economics and reliable operation across all hours and load levels. The best resource mix will have a balance of baseload, intermediate and peaking generation resources. The balance is needed for the system to work effectively in the differing parts of the load duration curve. Typically baseload plants have higher fixed costs and lower incremental costs, so they are operated around the clock. Peaking resources typically cost less but have high incremental costs. If you don’t have to run them very often the extra incremental fuel costs are made up for by the savings in fixed costs. Intermediate plants take the middle ground. Generally it costs more per installed MW for a more efficient plant. The more a plant is expected to operate the more those dollars can be justified. For illustration purposes the chart below illustrates hypothetical cost distributions for potential peaking, intermediate and base load plants based on the percentage of the time they operate (capacity factor).
The average cost per KWH varies for each resource type by capacity factor. In this hypothetical example it would not make sense to compare peaking, intermediate and baseload resources at any single specified capacity factor. It would make even less sense to compare their average costs when they areoperating at differing capacity factors.
Depending on the system they will supplement, any of the three might be the best choice. Put simply if you have a lot of low variable cost generation sitting idle, peakingresources will likely be justified. If high variable cost generation is running too frequently, your next additions should likely be baseload or intermediate. For this chart, if the new resource is going to run at less than 20% of its capability, a peaking plant will provide the best economics. If the plant needs to run full load around the clock, investing in baseload generation makes sense. At moderate levels of operation the intermediate resource provide better economics.
Traditionally generation was added to meet peak demand (left side of the load duration curve), but theresource type was selected to optimize energy costs across all hours of the curve in concert with existing generation mix.
The above tools give help us sort out options, but don’t have sufficient complexity to evaluate the multiple factors affecting economics over a generating plants life. After looking at the options with the above tools, the potential “best” generation additions are identified through detailed computer modelling. The models included system hourly loads, existing generation costs, fuel costs, operating limits and capabilities (ramp rates, heat rates…), maintenance, outages and so on extended out as muchas 30 years out into the future. This is done on an hourly basis across multiple scenarios varying factors such as fuel cost escalations, regulations changes, the availability to purchase or sell energy to other system and alternative growth scenarios. Generation resources are typically selected to operate the most economically with minimal risk across the range of scenarios judged most credible. The goal is to achieve reasonable costs, hedge risks and provide flexibility. The selected plants may not be the optimal choice in any given scenario, but rather the selected plants are “best” because they perform well across a diverse mix of potential scenarios.
Operating the System
As the future often doesn’t correspond to our best guesses, generation often does not operate as projected. Fuel costs change and the actual operation of plants may be much different from what was forecast. When dispatching plants, sunk costs are ignored and dispatch is driven by incremental costs.
Power systems operation can depart from planning expectations. For example, natural gas costs now are much lower than in many past forecast and costs associated with coal generation have risen above projected levels. Thus in many areas coal plants are functioning as intermediate plants, while combined cycle plants natural gas plants function as base load units. Alternative scenarios studied through planning models will mimic these sorts of transitions in future years as new resources are added and alternative fuel projections are modelled.
Providing Value to the System
Of prime importance is the ability of proposed resource is its ability to increase the system capacity so that peak load requirements can be met. Ideally you will increase the system capacity with resources that balance fixed costs and incremental costs in an appropriate manner for the system demands and existing resource mix. The value of any particular addition will be highly dependent on the existing resource mix in a given area. There are times and areas where baseload additions are needed and times and areas where peaking or intermediate generation is needed. Finally generation resources can add value through the provision of system services, such as load balancing, var support, inertia and frequency regulation.
Below is a sample table for various resource types highlighting the value they bring to a power system. Some of the characterizations are vague and most could be argued. As this is a “101” level positing, I’m not trying to resolve the issue of what resources are best, but rather setting up the framework for how they should be compared. Greater specificity would be warranted when looking at real choices for actual areas. This is a conceptual presentation and for space reasons I limited my column selection to major drivers and factors which tend to be overlooked. However, the table below could be expanded to include an additional columns showing “fuel escalation risk” and “carbon impacts” which would highlight benefits of alternative generation. In the final analyses we need to consider all significant factors.
Fitting Wind and Solar in with Traditional Generation Planning
The greatest challenges in fitting wind and solar resources within the traditional generation planning framework revolve around their limited ability to increase system capacity and provide support to meet system peak demand load levels. Intermittent resources are not dependable on the left side of the load duration curve.
Various formulas and approaches have been taken for deriving a “firm” capability contribution based on the installed wind or solar capability. At the high end, solar resources with tracking capability can in some areas approach capacity values of 75% of installed capacity. Near the lower end ERCOT derates wind to 8.7% of installed capacity. For systems with a winter morning peak, solar may need to be derated to near 0.
Assuming an intermittent alternative where a 30% derate would provide sufficient assurances of on-peak availability you need you would need an installed capacity value more than 3 times as high as the target for conventional technology. Another alternative would be to use the target addition for the renewable resource and then add an additional 70% of the target value in backup combustion turbine resources. The savings from the low incremental cost wind and solar resources however, are not usuallysufficient to justify the large construction cost for so many additional MW of capability (but not on-peak capacity). The ability to serve peak load and increase system capacity has been essential for project justification in traditional planning justifications.
The next challenge in justifying solar and wind projects is that their intermittent operation cannot be concentrated in the left hand side of the load duration curve for maximal economic and reliability benefits. Comparing the average cost of wind to the average cost of another resource, ignores the fact that the other resource can generate almost exclusively as needed in the left most part (where displaced energy prices are the highest) of the load duration curve. Even when intermittent resources generate on the left side of the chart, they do not have the same value as other concurrently operating resources because the intermittent resources cannot be fully counted on as being dependable.
On the right side of the load duration curve intermittent generation can cause problems. While generally it is desirable to back down costly generation and replace it with renewable resources, that is not always the case. There are times when generators must be kept on-line to meet upcoming daily peaks or avoid costly shutdowns and startups. Operators struggle at times backing down plants to minimum operating levels to make sure they have enough capable resources for the coming daily peaks. Intermittent energy generated during these periods can impose negative costs on the system. It makes sense for renewables to operate when they are displacing more costly resources, but not when they are contributing to system problems. Such counter-productive generation however, works to lower the average cost for solar and wind.
Between Average Cost and Detailed Studies
For completeness I should mention that there are other methods to compare generation resources that are more sophisticated than average cost and less cumbersome than detailed area by area detailed analyses. Levelized Cost of Electricity (LCOE) was developed to allow differing competing technologies to be more easily compared. While LCOE is a better measure than average cost, it is vulnerable to serious criticism as to its appropriateness for comparing dispatchable and intermittent resources.
I’ll leave it to those readers who may be interested to investigate further. See the references via hyperlinks provided earlier and here, here, here and here. My perspective is that measures such as LCOE are flawed but may be suitable to encourage the consideration of options and drive study work. However by themselves they can be misleading and do not give sufficient system specific guidance to support the adoption of programs, mandates or targets on their own. In any case the results of detailed modeling should be expected to be more accurate and appropriate.
Determining the value of various potential generating resources is significantly more complicated than comparing the costs of MW production by various technologies. Average cost comparisons can be very misleading. Electricity value varies by time of day, whether it can be scheduled or not, how it ties to other resources and a host of factors which cannot be accounted for in “average cost” calculations. Itmatters very much when energy is generated, where the energy comes from and how well it works to support the system. The average cost of solar and wind will need to be significantly below the average cost of conventional generation well before wind and solar can begin to approach general competitiveness on a cost basis.
Traditional generation planning studies can and have been run with various future renewable scenarios.As opposed to comparing “average costs” these studies can provide reasonable expectations for system costs and performance. I do not know of any detailed studies (even with highly optimistic assumptions around renewables and pessimistic assumptions for conventional technology) showing competitive costs from the the large scale integration of today’s renewable resources into any US power system. To date, most such studies overwhelmingly show significantly higher costs associated with high penetration levels from wind or solar resources. (Please share if you know where I can find studies that show widespread intermittent penetration scenarios that have feasible cost differences.)
We need more dialogue on the true costs associated with increased renewables. When renewable resources are mandated and the detailed published studies only examine “compliant” alternatives and scenarios, the true “extra” costs of renewables will be hidden. Without proper detailed alternative studies, we can’t really begin to estimate the costs of a transition to renewables. Projections, based on average costs, that suggest we can transition to renewables at reasonable costs are worthless if they do not address the many ways in which MWs are not equal.
JC note: Planning Engineer has posted two prior essays at Climate Etc.
As with all guest posts, please keep your comments relevant and civil.