Keep calm and calibrate

Sam Shamsi


Model calibration has become increasingly more challenging as collection system models have increased in size and complexity. Rigid misinterpretation of numerical model calibration criteria as a model pass/fail test rather than a QA/QC measure is causing expensive project delays.

This paper will present new methods and innovative ideas to overcome model calibration challenges which are expected to be particularly useful for mega models with more than 25,000 nodes or links.

An overview of model calibration criteria will be provided from various sources including US Environmental Protection Agency (EPA), Water Environment Federation (WEF), and United Kingdom’s (UK) Chartered Institution of Water and Environmental Management (CIWEM). The criteria will be reviewed based on their application to current collection system modeling paradigms such as mega models, continuous simulation, distributed modeling, water quality modeling, and 1D/2D modeling. Needs for updating the existing criteria will also be identified.

A popular source of numerical model calibration criteria is UK’s Chartered Institution of Water and Environmental Management (CIWEM). This criteria was first published in 1993 and last revised in 2002 in CIWEM’s Wastewater Planning Users Group (WaPUG) Code of Practice for the Hydraulic Modeling of Sewer Systems. This document which provides only event-based model calibration criteria was revised in 2017. The paper will summarize the changes in the new edition.

EPA provided model calibration guidelines in a 1999 CSO Guidance Document. This document provides only event-based model calibration guidelines rather than a WaPUG like numerical criteria. The key points of EPA guidelines will be presented. 

Model calibration alone is not sufficient to assess the accuracy of a mega model. Excessive focus on model calibration can result in neglecting other important factors that can impact the accuracy of mega models, such as model resolution, connectivity, invalid (out of range) attributes,  negative slope pipes, dry pipes, and numerical stability. This undesirable situation can result in treating a model as a black box which results in model mis-calibratiion. Models should be, therefore, audited for other sources of error and uncertainty first and foremost before attempting calibration. It is almost impossible to detect all the input errors in mega models without automatic model auditing tools. This section will provide model audit criteria for this purpose.

 Click here to download a static PDF version of the presentation.

Permanent link: