# hydraulic calibration

Post on 07-Nov-2014

85 views

Category:

## Documents

Embed Size (px)

TRANSCRIPT

AWWA MANUAL

M32

Chapter

4 Hydraulic Calibration

4.1. INTRODUCTION _______________________________________Hydraulic models are only as good as the accuracy of their predictions of actual water system performance. The best way to verify that a hydraulic model provides results that closely match actual eld conditions over a wide range of operating conditions is to perform model calibration. This chapter will provide guidance on the steps required to calibrate a hydraulic model, describe the various types of calibration, and describe the issues and pitfalls to keep in mind when conducting a model calibration. Hydraulic models are essentially mathematical representations of the water system. They must contain accurate data dening the water usage in the system, the physical characteristics of the piping and facilities in the system, and the control settings of equipment and devices needed to replicate accurate system performance. A model may complete a simulation based on the information provided, but, if this information does not accurately reect the system being analyzed, the result will have limited benet and may actually lead to erroneous conclusions. Calibration can be tedious and time consuming. However, the potential benets of a well-calibrated model for making accurate engineering decisions far outweigh the potential cost of mistakes caused by incorrect model results.

4.2. WHAT IS CALIBRATION? ________________________________Calibration is the process of comparing the results of model simulations to actual eld data and making corrections and adjustments to the model to achieve close agreement between computer-predicted values and eld measurements. Typical comparison values include pressures, ow rates, and reservoir water levels. A well-calibrated model will stay in close agreement with eld measurement over a wide range of operating conditions. Discrepancies between model results and eld data can be due to many factors including errors in model construction, errors in reported eld data, unknown constraints or controls in the water distribution system, and inaccurate assignment of model variables. Some of the parameters related to physical facilities that may require correction during calibration include system connectivity, pipeline diameters, node

85

86 COMPUTER MODELING OF WATER DISTRIBUTION SYSTEMS

elevations, control valve settings, check valve direction, and pump characteristics. Some of the more variable or estimated data that may require adjustment include assignment of minor losses, pipe roughness coefcients, and demand peaking factors. It is important to remember that eld data include a certain level of uncertainty and is too often inaccurate or simply wrong. It should not be immediately assumed that there is something wrong with the model if it is not able to match the observed data. Inaccuracies should be investigated carefully.

4.2.1. Calibration, Reconciliation, and ValidationThe term model calibration is deemed by some to be a misnomer because it implies that the model will be adjusted to match the actual system. However, in many cases, there are situations in the system that cannot be simulated in the model, such as unknown closed valves. For this reason, some suggest that the term model reconciliation is a better term than model calibration. For purposes of this manual, the generally accepted historical term calibration will be used. It is acknowledged that the degree of calibration is highly dependent on the accuracy of available data and potential unknowns in the distribution system. The term reconciliation acknowledges that when model results do not match recorded data, the recorded eld data may be incorrect or there may be unknown anomalies in the distribution system, such as closed valves. Understanding that a distribution system model has inherent limitations can prevent inappropriate use of the model. Adjusting input data to achieve a higher degree of calibration or best-t between modeled and measured results, without sufcient justication, will not negate the inherent limitations in the data. The frequent use of a model, along with a high degree of familiarity with the system, is the best method for establishing model validity and will help the user to understand the limitations of the calibrated model. Validation is a term often used to refer to the process of checking the results of a hydraulic model following an update process. Validation may also be used to refer to comparing the model to a different set of eld data than that for which the model was calibrated. For example, calibrating on August 12 and then validating against conditions on August 14. Validation is considered by some to be not as formal as the calibration process and may involve a check of results to verify that pressures and ows are within expected and accepted ranges. The term validation is used in many ways by different modelers and is not as well dened a term as calibration.

4.2.2. Reconciling DiscrepanciesModel results and measured values should be compiled into tables and graphs. The difference between measured and predicted values should be analyzed to see how closely the model reects the actual system. If many of the differences are great, something may be wrong with the model setup or there may be errors with connectivity. When only a few differences are signicant, the modeler should concentrate on the areas where large differences exist. Reconciling differences at this time requires patience and ingenuity. Sometimes tweaking the model will bring simulated and measured results closer; sometimes talking with operators will reveal additional information that was not included in the original model. Calibration is an iterative process throughout which the modeler must carefully make assumptions, because calibration is not just about minimizing differences but more about getting the model to correctly represent the system. Unknown conditions in the distribution system can often result in calibration inaccuracies. Any model assumptions made in the adjustment process should be the result of clear and logical ndings in the system. For example, a large leak or a

HYDRAULIC CALIBRATION

87

mistakenly closed valve may cause pressures discrepancies, which should be corrected in the eld and replicated in the adjusted model. A key point of model calibration is a full understanding why the model and eld data do not agree. Once these differences are understood, adjustments are easier to identify and apply.

4.2.3. Types of CalibrationModel calibration can be categorized several ways as summarized below: s Hydraulic versus water quality simulation s Steady-state versus extended-period simulation s Macro- versus microcalibration Currently, most models are calibrated only to hydraulic parameters to verify that pressures, ows, and water levels can be accurately simulated. However, calibration to water quality parameters is becoming more common. Two common water quality calibration parameters include chlorine residuals and disinfection by-products (DBPs). Water quality calibration requires additional data to be input into the model; for example, chlorine calibration requires input of initial chlorine concentrations, chlorine bulk decay curves, and wall decay coefcients. Further discussion of water quality calibration can be found in chapter 7. Steady-state calibrations are by their nature purely hydraulic calibrations. They simulate distribution system hydraulic performance at a given moment in time, usually a high demand condition such as maximum hour or hydrant ow. Extended-period simulation (EPS) calibrations can be either hydraulic or water quality calibrations. EPS analyses are a series of steady-state simulations linked together to approximate the behavior of a system over a period of time. EPS analyses model changes in demands, operational controls for pumps and valves, and storage facility water levels. EPS analyses are required for water quality simulations and should be conducted using a model that has been hydraulically calibrated under EPS conditions. Macrocalibration generally refers to calibrating the entire system or an entire pressure zone to recorded data captured by recording charts and/or a SCADA system. Macrocalibration focuses on large discrepancies between observed and predicted values and helps identify gross model errors, such as erroneous model connectivity, nodal elevations, or operational settings. Microcalibration generally refers to analyses that verify localized conditions in a specic area are being accurately simulated. Microcalibration can be equated to netuning the model in a specic area. A typical microcalibration is the use of simulating recorded re hydrant ows to verify the correctness of localized piping and valve closures. In smaller systems and in systems with minimal historical data capture, the use of calibration to re hydrant test results may be used as the basis for macrocalibration of the model. There is usually an overall macrocalibration performed when a model is rst constructed or updated. However, if a model is used for a specic design study, it is often desirable to investigate the area in question and conduct a microcalibration with a re hydrant ow test as a check of model accuracies in the study area.

4.2.4. Calibration GoalsThe goal of calibration is guided by the model end use. A model calibrated for master planning, for instance, may not be calibrated sufciently for water quality analysis. The real guideline for determining if a model is calibrated appropriately is whether