skip to main content
Conceptual Overview
USACE‑SWD Modeling Techniques
Conceptual Overview
This document describes functionality developed to replicate algorithms used by the U.S. Army Corp of Engineers Southwest Division (USACE‑SWD). These algorithms, consisting of object methods, RPL predefined functions, and other utilities were developed as a replacement tool for the USACE‑SWD FORTRAN SUPER program. Although the algorithms were implemented for the USACE‑SWD, they are general in nature and available to all RiverWare users.
The purpose of this document is to consolidate all of the documentation specifically developed and delivered to the USACE‑SWD to describe the algorithms. This document contains documentation and implementation Help specific to the USACE‑SWD model setup and approach. But it contains links to the rest of the RiverWare Help to describe the general methods and functions. The document first presents an overview of the approach used. Then it describes the steps necessary to set up a basic simulation model and run it as a calibration. Next, there is a description of the process used to define operating policy using a RPL ruleset. Each USACE‑SWD policy is then described including the necessary method selections, how to define the policy in RPL, and detailed description of the algorithm. Finally, sections are provided to describe how to set up a yield study, use the Data Management Interface, use output and analysis tools, and how to improve performance of the modeling runs.
This document is intended to guide a user from a blank workspace through adding objects, implementing policy, and finally implementing a yield study.
The integration of USACE‑SWD algorithms in RiverWare brings together the RiverWare object-oriented modeling features, the power and flexibility of the priority rulebased simulation, and the Computational Subbasin, that allows the execution of multi-object computations from a rule function. The algorithms are thus implemented in modular, object-specific contexts for ease of maintenance and extension, as well as flexibility of use through user selectable methods. Following is a brief overview of the modeling steps and components.
Modeling Steps
Figure 1.1 shows the typical modeling steps to implement the functionality described in this document. An overview of each step is described in the following sections and the remainder of this document describes these items in detail.
Figure 1.1  Modeling steps diagram
Develop Simulation Model
The first step is to develop a simulation model of the system by pulling objects off the palette, selecting the appropriate methods, linking objects together and specifying required data. Next, timeseries data is specified. The data management interface (DMI) can be used to import data as desired. With enough data, the model will run. At any point, the output utilities can be used to analyze and view results. The model should be calibrated to make sure that the inputs are producing the desired outputs. For example, you may input inflows to the system and reservoir outflows. You would then make sure that reservoir pool elevation or storage are correct and that downstream flows are routed correctly.
Implement Policies Using Rulebased Simulation
The model can then be converted to a predictive mode by removing some of the inputs (E.g. Storage, Pool Elevation, and Outflows) to under-determine it. Then, rules can be implemented to compute the reservoir outflows based on the state of the system and the operating policies.
Described in this document are the following policies: surcharge, regulation discharge, flood control, low-flow releases, reservoir diversions, hydropower releases, and fish releases. Each policy may have additional methods to be selected and data to be input. Then, rules can be written in the RiverWare Policy Language (RPL). Many of the rules use specific predefined functions that execute methods on objects.
Run Using Unregulated Conditions
Sometimes, it is desired to run the model and compute the flows that would have occurred without one or more reservoirs in place. To do so, use scripts to change the model to use unregulated conditions. Use a script to make a run and generate snapshots of desired slots.
See “Modeling Unregulated Conditions” for details on defining and running an unregulated system.
View and Analyze Results
Given the model and the ruleset, a run is made. There are numerous tools and utilities that can be used to view and analyze results including the following:
• Plotting: present a graphical view of the data.
• Statistical slots: create frequency duration or exceedence curves. Then, plot them on a probability scale.
• Expression slots: create user defined expressions involving slot or other values. For example, this can be used for aggregation of data.
• Data Management Interface: DMIs can be used to export data to an external sink.
• System Control Table: SCTs can be used to view and edit data in a spreadsheet like format.
See “Analysis and Output Tools” for details on viewing and Analyzing results.
Convert to Yield Study, If Desired
A yield study is used to determine the largest constant diversion that can be made from a single reservoir such that the reservoir pool elevation will not drop down below the bottom of conservation pool at any time during the run period. Other operating policies like Surcharge, Regulation Discharge, Flood Control, Low-flow Release, and Hydropower are included in this analysis.
The Yield study uses the iterative Multiple Run Management configuration. In this configuration, the model is run multiple times with a different trial value for the diversion from a reservoir. After each run, logic is used to determine if the yield has been met and if not, a new trial value is calculated and a new run is made. Multiple reservoir can be included; each reservoir’s yield is computed separately, starting at the top of the system. Once the first reservoir’s yield is found, the next downstream reservoir’s yield is found. Thus, the model is run multiple times to find the yield from one reservoir, then that reservoir’s yield is locked in and the analysis is repeated on the next reservoir.
An existing model is converted to a yield study by importing data and an MRM RPL set. Then, the user sets up an MRM configuration and specifies the search algorithm (Bisection or a heuristic approach). The user can also specify the convergence criteria including the initial trial values and the maximum iterations. Finally, the run is made. The SCT can then be used to view the results of the multiple runs. Included in the results (for each reservoir) is the yield, the critical draw down date, the critical draw down period and the critical period duration.
The conversion to a yield study is a separate modeling step that may or may not be required for a given basin or portion of a basin. See “Computing Reservoir Yield” for complete details.
Model Components
Implementation of the USACE‑SWD functionality makes use of many of the RiverWare modeling features. Following is a description of the various pieces.
1. Methods on objects
On reservoirs, there are user selectable methods to specify how to calculate the incremental flows, forecast hydrology, surcharge release (mandatory releases), flood control, low-flow releases, reservoir diversions, and hydropower method.
Control Point objects are used in the basin network to represent channel control points that influence flood control and other releases. User selectable methods on Control Point objects support the calculation of forecasted inflows (uncontrolled area flows), regulation discharge, and empty space hydrographs. The user selectable Key Control Point Balancing method allows for the computation of a balance level for associated reservoirs.
The reach, water user, and diversion object also have user selectable methods that are used as part of the USACE‑SWD functionality.
2. Computational Subbasin
The RiverWare computational subbasin has object attributes including user selectable method categories, methods, and data slots to support calculations that involve multiple simulation objects simultaneously. The flood control category includes two methods that govern the nature of the flood control solution. The Operating Level Balancing method is used to model flood control based on the USACE‑SWD algorithms. The Phase Balancing method is based on the USACE-KC flood control operations. This document deals with the Operating Level Balancing calculations. The Phase Balancing method will not be discussed in this document. In addition to the flood control category, several related categories and methods are available to adjust/modify some of the details of the flood control algorithm.
Additionally, there are user selectable methods to initialize objects within the subbasin, compute incremental flows, computer reservoir diversions, compute low-flow releases, and compute alternative routing coefficients within the subbasin.
3. Surcharge Release and Regulation Discharge Flags
The surcharge release and regulation discharge flags are used to trigger the surcharge release and regulation discharge calculations on reservoirs and control points, respectively. Rules are used to set the surcharge release flag on the outflow slot of each reservoir. When this flag is set, it triggers the reservoir to dispatch and compute a surcharge release forecast. The regulation discharge flag is set on all control points simultaneously by a single rule. This flag triggers each control point to solve for its regulation discharge and empty space hydrographs according to the user methods selected on each control point object.
4. Rule functions to implement the policy
Rules execute predefined rule functions that implement the policy. For example, a predefined rule function called FloodControl invokes the computational subbasin method and returns the solution by way of a list of object.slot and value pairs to be set by the rule. The FloodControl function executes the currently selected method in the Flood Control category on the computational subbasin object. The method is executed on the subbasin named in an argument passed to the FloodControl rule function.
Additionally, the following predefined functions execute the policy listed:
– MeetLowFlowRequirement. Computes reservoir release to meet a demand at a control point
– ComputeReservoirDiversions. Computes the diversion from each reservoir to meet water user demands.
– HydropowerRelease. Computes the additional release required to meet the specified load without causing additional downstream flooding.
Ruleset and Policy
The ruleset implements the policy that is used to make reservoir releases and diversions from the system. The following policies apply to one or more USACE‑SWD applications. Presented is a conceptual overview. Each piece is described in detail in the document.
Surcharge Release Rules
The surcharge release calculations are executed by the reservoir object when the surcharge release flag is set on the outflow slot by a rule. Each rule sets the surcharge release flag on the outflow slot of a single reservoir object. The rules must start at the upstream end of the basin. When the surcharge release flag is set, the reservoir will dispatch during post-rule simulation if the Inflow to the reservoir is also known. During dispatching, the reservoir will compute the surcharge release forecast and set the Outflow slot for the current timestep and all other timesteps in the forecast period. These outflow values propagate via links to downstream objects and are routed using the selected routing method on any Reach objects that are encountered. When the flow values reach a downstream reservoir, that reservoir will have Inflow slot values for the forecast period. A subsequent rule will set the surcharge release flag on that reservoir resulting in dispatching. In this way, each subsequent downstream reservoir calculates the surcharge releases for the forecast period considering the routed surcharge releases from upstream reservoirs.
Flood Control Minimum Release
Using rule logic, Flood Control Minimum Release values can be set after the surcharge operation. It is up to the user to define the rule logic for these releases. These releases are considered minimums because they are made regardless of what the flood control logic computes. Once set, the releases are allowed to propagate downstream and occupy channel space.
Regulation Discharge Rule
After the above rules have executed and the releases have been routed downstream, a single rule sets the regulation discharge flag on the Reg Discharge Calculation slot for all Control Point objects. This triggers each control point to dispatch and execute the selected regulation discharge methods. Because the uncontrolled area inflow forecasts have already been computed and the surcharge releases have been routed downstream, each control point contains the total discharge at its location for each timestep in the forecast period. This information is used to appropriately compute the regulation discharge and empty space hydrographs.
Flood Control Release Rule
After the regulation discharge rule has executed, the rule to calculate the flood control releases can be executed. This rule calls the predefined FloodControl function which executes the selected Flood Control method on the Computational Subbasin. The FloodControl function returns a list of flood control releases that should be made for each reservoir at the current timestep. Also included is the outflow from each reservoir at the current timestep. The outflow is computed as the flood control release plus the surcharge release plus the flood control minimum release. The flood control rule sets the Outflow slot and Flood Control Release slot on each reservoir, at the current timestep, given the values returned by the FloodControl function. When these slots are set, it triggers each reservoir to redispatch using the new Outflow value. The reservoir objects solve, compute new storage and pool elevation values (as well as execute any user selectable methods), and the new Outflow values are routed downstream.
Low-flow Releases
Low-flow demands are specified at control points and each control point has a list of reservoirs that are used to meet a given demand. Also, each reservoir has a specified maximum delivery rate for meeting a low-flow requirement.
Once flood control is complete, then the Low-flow rule executes and calls the MeetLowFlowRequirements predefined function. This function determines the low-flow release for each reservoir serving a low-flow demand (i.e. a control point) as follows: If there is a low-flow shortage, the serving reservoirs are sorted by level in descending order. Each reservoir (beginning with the most full reservoir) makes a release until the requirement is met, the maximum low-flow release on the reservoir is met, or the reservoir reaches the bottom of the conservation pool (whichever value is lowest).
When the reservoir Outflows are set, the system resolves to route the flows downstream. The low-flow shortages will be recomputed to reflect the low-flow releases. The reservoir Operating Levels will be recomputed to reflect the new releases. Then the next rule will execute to determine the low-flow releases for the next control point.
Reservoir Diversion
Once low-flow releases have been made for all reservoirs, direct from reservoir diversions are calculated by calling the predefined function ComputeReservoirDiversions. Reservoir demands can be a water user or another reservoir. Each demand can draw from multiple reservoirs and each reservoir can act as a source for multiple water users or demand points.
The diversions from the reservoir are modeled using diversion objects while the actual demands are modeled with water user objects that compute their Diversion Request (water supply requirement). Even if the demand point is another reservoir, the demand is modeled with a Water User object (configured to have a 100% return flow which could then be sent to the demand reservoir).
When called, the ComputeReservoirDiversions function computes the diversion to each water user by visiting each reservoir (in order of highest level first) and trying to meet the diversion request but limited by maximum delivery amounts, not drawing below the conservation pool. Also, if the demand is a based on reservoir level, no diversions are made if the demand reservoir and has a higher level than the supply reservoir, or the demand reservoir is in the flood pool.
The rule sets the Supply From Reservoirs and the Incoming Available Water slots causing the water user object to solve. It then propagates values to the diversion object and the reservoir which also dispatch and solve.
After surcharge releases, flood control, low-flow releases, and reservoir diversions have been made, a reservoir may release additional water from its power pool to meet a specified power commitment, i.e., the Load specified for that reservoir. The hydropower rule executes the HydropowerRelease predefined function. This function tries to meet the specified or computed load but is limited to prevent the following conditions:
• The pool elevation from dropping below the minimum power pool elevation
• Releasing more than generating capacity
• The pool elevation from dropping more than the maximum allowable power pool draw down
• Additional downstream flooding
The rule sets the additional hydropower release and increases the reservoir Outflow. This causes the reservoir to redispatch and propagates the water downstream.
Fish Release
On a power reservoir, during the hours of the day when power is not generated, water may be released to meet minimum fish flows. These are called Fish Releases or QFish releases. The release of water to meet a downstream demand is determined based on the available fish water in the reservoir, the other off peak releases that are occurring and the monthly minimum fish flow demand. The fish storage in the reservoir is accounted for each timestep. Fish water is either released or is lost due to evaporation. It is replenished through the following methods: all inflow is shared, the accumulated local inflow is shared, or no inflow is shared. Also, anytime the pool exceeds the top of conservation pool, the fish storage is replenished.
Rule-Simulation Interaction for Each Timestep
To summarize, the following are the steps that are taken for each timestep in the model.
1. At the beginning of the timestep, cumulative local inflows are disaggregated spatially and an inflow forecast is computed for each reservoir and control point object.
2. The surcharge release rules execute in upstream to downstream order and set the surcharge release flag on the Outflow slot of each reservoir. After each rule executes, the affected reservoir dispatches and computes its surcharge release forecast. These releases are routed downstream to the Inflow slot of the next reservoir before the next rule executes.
3. The flood control minimum release rules optionally set the reservoir outflow and flood control minimum release based on the rule logic. After each rule executes, the affected reservoir dispatches. These releases are routed downstream before the next rule executes.
4. The regulation discharge rule executes and sets the regulation discharge flag on the Reg Discharge Calculation slot on each control point. The control points dispatch and compute regulation discharge and empty space hydrographs for the forecast period.
5. The flood control rule executes and invokes the FloodControl predefined function. This function computes the flood control releases for the subbasin and returns the flood control release and outflow values, for the current timestep, for each reservoir. The flood control rule sets the flood control release and outflow values on each reservoir. The reservoir objects redispatch and solve for pool elevation, storage and any other reservoir methods that have been selected. The outflow values are routed downstream.
6. For each Low-flow Release control point:
– The low-flow release rule execute and invoke the MeetLowFlowRequirements predefined function. This function computes the release from each reservoir in the subbasin to meet the low-flow demand at the control point. The rule sets the Low-flow Release and Outflow slots causing the objects to dispatch and propagate the effect.
– The low-flow diversion rules execute to route any low-flow releases around lagging reaches by setting Diversion Requests equal to the low-flow release. This causes the reaches and water users to redispatch. Thus, low-flow releases arrive at their control point destinations immediately.
7. The reservoir diversion rules execute and invoke the ComputeReservoirDiversions predefined function. This function computes the diversion that must be made from each reservoir in the subbasin to meet the demands in the subbasin.
Note:  A single reservoir can serve multiple demands and a demand can be met from multiple reservoirs. After setting the diversions, the reservoir, diversion objects and water users redispatch.
8. The hydropower rule executes the HydropowerRelease predefined function. This function executes the hydropower release function on the flood control subbasin to determine the additional hydropower release that is necessary to meet the load without causing additional downstream flooding. The rule sets the Additional Hydropower Release slot and resets the reservoir Outflow slot. This causes the system to redispatch and solve the entire system.
9. The Fish Release rules (where defined) execute and determine the additional outflow to meet the Fish Release demand. The rule sets the reservoir Outflow and reservoir Off Peak Spill slots. This again causes the system to redispatch and solve. The rule also performs accounting actions to calculate and set the inflows and outflows to the fish account on the reservoir.
10. If no other rules exist, the timestep is complete. RiverWare moves to the next controller timestep and the process is repeated.
Revised: 06/03/2019