Water Supply Planning - Reservoir Operations Models
Reservoir models at their core are simply mass balance equations, just like balancing a check book. Reservoir models can range from simple monthly models with one or two variables, all the way up to models with multiple linked reservoirs with 50 or more variables calculating daily inputs and outputs. The level of complexity of the model will match the level of complexity of the reservoir project being modeled.
Simple models are often run on monthly time steps with limited inputs and outputs, minimal variables, and are most suited for river systems with consistent flows, and consistent demands. Models for irrigation reservoirs can often be run with monthly time steps and can provide representative results with reduced cost and complexity. These monthly time step models are better suited to modeling for a company or water user to better understand their practices, but where the model will be presented as evidence in Water Court cases a more refined time step is necessary.
As additional complexity is added to the model, or is being used to support a water rights application, it is more advantageous to switch to a model with daily time steps if daily data for the inputs are available. That complexity can be introduced to the model in many ways. Significant daily changes in flow availability to the reservoir, multiple reservoirs in the same model, models which need to be tested to determine the most efficient operating strategies, and models which will be presented as evidence in Water Court are all examples of models which likely require daily time steps.
Models are most effective when they give you more than binary results, e.g., did the model meet the demand, yes or no? That is not a terribly helpful result. Models with more built-in variables allow for sensitivity analyses to be easily performed on any one, or multiple, variables to estimate the effect of important parameters like the size of the reservoir, the rate of diversion, or rate of release from the reservoir. This is especially important in reservoir models intended to predict future project yields. In-depth modeling and sensitivity analyses with good data can highlight areas where infrastructure can be built smaller while providing the same desired result, saving significant sums of money on construction costs.
At LWS we tend to make parameters variable in a model control panel instead of fixed parameters in the model formulas so the model can provide more flexibility in the analysis phase. We also build these models in Excel to provide clients with a model they can easily see all inputs and outputs. This prevents the model from becoming a black box. Using a control panel of variables in the Excel model allows us to perform easy-to-run sensitivity analyses on more variables, which gives the client a more fine-tuned and efficient reservoir system, and allows us as engineers to develop a better understanding of the chokepoints in the system which may not have been previously obvious.
These variables also allow us to make rapid changes to the Excel model and provide updated results to the client in real-time during planning sessions, whereas changing fixed variables in black box models often require engineers to make changes at a later time and report back with updated results. For example, in roughly 30 seconds we can use our Excel-based reservoir model to identify the minimum headgate flow rate needed to meet the required yields. It takes the Excel model about one minute to generate a table of yields for all the potential headgate flow rates. This analysis can be performed with any variable and output combination in the model. Since unforeseen changes to the project parameters are nearly always a factor, it’s a good practice to make parameters variable from the start and to build the model in Excel.
The final cornerstone of a good reservoir model is making individual formulas as simple as possible, especially in large and complex models. While it may seem like a time saver to make a model in as few steps as possible, it makes changes to that model down the road terribly difficult, while also making the QA/QC process harder. While it may seem counter-intuitive, having a model with more lines of simple formulas also runs faster than a model with less lines of complex formulas.
Using these principles, LWS has generated highly-complex models with multiple-linked reservoirs in a system that provides daily results for 20 or more years by running more than a half million individual calculations but can produce updated results in less than five seconds when parameters are changed. Models like this are an invaluable tool during strategy sessions. Where previously, it may have required multiple meetings to come up with potential ideas to improve a reservoir system and then have the engineer test those ideas and report back; with a LWS model these ideas can be tested in real-time during strategy sessions where everyone can see and follow along while viewing changes as they occur in the Excel model. This allows us to dramatically shorten the time and expense to move the project to the next phase.
For even more functionality, these reservoir operational models can be integrated with point flow models of streams. Point flow models are models which estimate flows in river systems at multiple locations in the system for each time step of the model. These models can be used to analyze reservoir project yields as well as the project’s impact on all the point flow locations in the model. For more information about point flow models, stay tuned for our upcoming blog on the subject!
If you have a new reservoir project in your organization’s future, reach out to LWS and our team will be happy to provide a free consultation and project proposal tailor made to your needs.
Chris Fehn – Senior Project Engineer – chris@lytlewater.com
Anna Elgqvist – Senior Engineer – anna@lytlewater.com
Bruce Lytle – President – bruce@lytlewater.com
