A Straightforward Approach to Generating One Set of RTK Parameters for Continuous Calibration

Hazem M. Gheith

ABSTRACT

Continuous simulation facilitates the evaluation of collection system surcharges and overflow and water in basement frequencies over annual or longer periods. Continuous simulation is important to both regulatory agencies and sanitary authorities who are increasingly interested in understanding collection system performance under a variety of storm levels and durations. Understanding the continuous performance is important to the agencies because it allows for quantifying frequency, volume, and duration of overflows, and understanding water quality impacts on receiving waters. For authorities, this understanding helps ensure that capital improvements for system mitigation are not overly conservative, thereby reducing costs and billing rates. Generating one set of RTK parameters, which successfully regenerates observed wet weather flow (WWF) responses from all monitored storm events will ensure confidence and more accurate results when applied to non-monitored periods and synthetic storms.

Traditionally, modelers evaluate a range of design storms by generating several calibrated hydrology models from monitored wet weather events. Based on the availability of flow data and the focus of the study, modelers select a number of historical storms representing different levels and durations to calibrate these models. There are some problems with this traditional method. First, the generation of these calibrated models is time consuming. Additionally, when studying a non-monitored storm event or use a design storm with a duration or level that does not match the calibrated models, the certainty of the results are unclear. In some cases, this uncertainty can be reduced by conducting more flow monitoring. However, even with more monitoring, it is not guaranteed that a storm event with the required level and duration will take place during the monitoring period. Accordingly, generating one constant set of RTK parameters became the focus. In some practices, a median synthetic set of RTK parameters is statistically generated from individually calibrated events. Other methods like Genetic Algorism Calibration are also used. It is understood that the resulting set of parameters may not necessarily match any single event parameters.

This presentation presents a straightforward approach to generating a constant set of RTK parameters which provides good WWF simulation for all monitored storm events. The process relies on selecting an appropriate storm event that will allow the RTK parameters to not be dependent on climate or antecedent moisture conditions. The modeler should select a monitored storm event which was preceded shortly by another storm. The first storm should be large enough to cancel out the effect of initial abstractions and bring the sewershed into a saturated condition. In addition, there should be a very short dry period between the two storms in order to avoid large recovery of the initial abstractions. This situation will ensure that most of the second storm’s rain is excess rainfall volume after all types of rain losses have taken place. A percentage of this excess rainfall will then enter the collection system through the collection system’s connections and defects. Generating the RTK set based on such a storm event allows the parameters to be almost exclusively a function of the collection system’s size and asset defects and not affected by initial abstractions. After generating the one RTK set of parameters that matches the storm event WWF response, the modeler should apply appropriate maximum storages and recovery rates for simulating the WWF response from the previous storm event. Then the same set of RTK parameters is applied to the entire monitored period to calibrate events of other months by adjusting the values of maximum storage and recovery rates for each month. Adjustments to maximum storages and recovery rates will accommodate for seasonal variations.

This method of generating a unified set of RTK parameters was successfully used to characterize WWF conditions in a tributary area in the City of Columbus. Thirty-four meters were used to monitor flow response from upstream sewersheds and 15 meters were placed at relief points. Most monitoring activities were conducted between July 2007 and June 15, 2008. Monitored tributary areas varied in size from less than ten acres to 305 acres. The discussed approach to developing RTK parameters was applied to all of the metered sewersheds and a single set of RTK parameters was generated for each monitored sewershed. The generated RTK sets accurately simulated and matched observed WWF responses in each monitored sewershed for the largest 16 monitored storm events.

The described approach for generating RTK parameters is straightforward and reduces calibration efforts. The generated one set of RTK parameters closely describes the collection system deterioration condition. It can be applied to non-monitored and synthetic design storms with confidence. It will ultimately lead to more accurate recommendations for capital improvement projects.


Permanent link: