Ahmed Mohamed AbdelSattar

Abstract Stormwater management wet ponds are generally very shallow and hence can significantly increase (about 5.4 °C on average in this study) runoff temperatures in summer months, which adversely affects receiving urban stream ecosystems. This study uses gene expression programming (GEP) and artificial neural networks (ANN) modeling techniques to advance our knowledge of the key factors governing thermal enrichment effects of stormwater ponds. The models developed in this study build upon and compliment the \{ANN\} model developed by Sabouri et al. (2013) that predicts the catchment event mean runoff temperature entering the pond as a function of event climatic and catchment characteristic parameters. The key factors that control pond outlet runoff temperature, include: (1) Upland Catchment Parameters (catchment drainage area and event mean runoff temperature inflow to the pond); (2) Climatic Parameters (rainfall depth, event mean air temperature, and pond initial water temperature); and (3) Pond Design Parameters (pond length-to-width ratio, pond surface area, pond average depth, and pond outlet depth). We used monitoring data for three summers from 2009 to 2011 in four stormwater management ponds, located in the cities of Guelph and Kitchener, Ontario, Canada to develop the models. The prediction uncertainties of the developed \{ANN\} and \{GEP\} models for the case study sites are around 0.4% and 1.7% of the median value. Sensitivity analysis of the trained models indicates that the thermal enrichment of the pond outlet runoff is inversely proportional to pond length-to-width ratio, pond outlet depth, and directly proportional to event runoff volume, event mean pond inflow runoff temperature, and pond initial water temperature.

Summary One of the most challenging modelling tasks in hydrology is prediction of the total suspended sediment particle size distribution (TSS–PSD) in stormwater runoff generated from exposed soil surfaces at active construction sites and surface mining operations. The main objective of this study is to employ gene expression programming (GEP) and artificial neural networks (ANN) to develop a new model with the ability to more accurately predict the TSS–PSD by taking advantage of both event-specific and site-specific factors in the model. To compile the data for this study, laboratory scale experiments using rainfall simulators were conducted on fourteen different soils to obtain TSS–PSD. This data is supplemented with field data from three construction sites in Ontario over a period of two years to capture the effect of transport and deposition within the site. The combined data sets provide a wide range of key overlooked site-specific and storm event-specific factors. Both parent soil and TSS–PSD in runoff are quantified by fitting each to a lognormal distribution. Compared to existing regression models, the developed model more accurately predicted the TSS–PSD using a more comprehensive list of key model input parameters. Employment of the new model will increase the efficiency of deployment of required best management practices, designed based on TSS–PSD, to minimize potential adverse effects of construction site runoff on aquatic life in the receiving watercourses.

Summary One of the most challenging modelling tasks in hydrology is prediction of the total suspended sediment particle size distribution (TSS–PSD) in stormwater runoff generated from exposed soil surfaces at active construction sites and surface mining operations. The main objective of this study is to employ gene expression programming (GEP) and artificial neural networks (ANN) to develop a new model with the ability to more accurately predict the TSS–PSD by taking advantage of both event-specific and site-specific factors in the model. To compile the data for this study, laboratory scale experiments using rainfall simulators were conducted on fourteen different soils to obtain TSS–PSD. This data is supplemented with field data from three construction sites in Ontario over a period of two years to capture the effect of transport and deposition within the site. The combined data sets provide a wide range of key overlooked site-specific and storm event-specific factors. Both parent soil and TSS–PSD in runoff are quantified by fitting each to a lognormal distribution. Compared to existing regression models, the developed model more accurately predicted the TSS–PSD using a more comprehensive list of key model input parameters. Employment of the new model will increase the efficiency of deployment of required best management practices, designed based on TSS–PSD, to minimize potential adverse effects of construction site runoff on aquatic life in the receiving watercourses.

An innovative predictive equation characterizing watermain failure timing developed from datasets of historical failures in Greater Toronto Area (GTA), Ontario, Canada is described. Gene expression programming (GEP) is used to develop empirical relations between the time to failure and control variables including protection methods for three types of pipes, Cast Iron, Ductile Iron, and Asbestos Cement. The developed GEP model has a correlation coefficient of 0.68, and with the advantage for predicting not only the time to first break, but also subsequent breaks. The prediction uncertainties of the developed GEP were 38 {%} of the median value for time to next failure. A parametric analysis is performed for further verification of the developed GEP model showing the relation is simple, yet effectively forecasts watermain timing to first failure and subsequent failures. Simulated failure scenarios indicate a return to high failure rates if cement mortar lining and cathodic protection are not extended to all candidate pipes in the distribution system.

This paper presents an explicit solution for the specific flow depths in partially filled pipes of circular cross-sectional area. Four depths encounter in most classical free-surface flow problems, namely the normal depth, the critical depth, the sequent depths from the specific momentum equation, and the alternate depths from the specific energy equation. This paper proposes new equations derived from dimensional analysis and gene expression programming to estimate directly those flow depths. The equations are examined over a wide range of flow and geometric conditions, providing satisfactory predictions when compared with the exact solution obtained from the governing hydraulic equations. Maximum error encountered in the critical and normal flow depths predictions is less than 1.25%, and maximum error encountered in the alternate and sequent flow depths predictions is less than 3.85%, which are acceptable in most hydraulic engineering practice. The equations are simple and would be useful in hydraulic engineering practice when quick and accurate estimates are needed of those depths, and can also be used to find the initial values for the flow depth in various theoretical and numerical schemes.

This paper presents an explicit solution for the specific flow depths in partially filled pipes of circular cross-sectional area. Four depths encounter in most classical free-surface flow problems, namely the normal depth, the critical depth, the sequent depths from the specific momentum equation, and the alternate depths from the specific energy equation. This paper proposes new equations derived from dimensional analysis and gene expression programming to estimate directly those flow depths. The equations are examined over a wide range of flow and geometric conditions, providing satisfactory predictions when compared with the exact solution obtained from the governing hydraulic equations. Maximum error encountered in the critical and normal flow depths predictions is less than 1.25%, and maximum error encountered in the alternate and sequent flow depths predictions is less than 3.85%, which are acceptable in most hydraulic engineering practice. The equations are simple and would be useful in hydraulic engineering practice when quick and accurate estimates are needed of those depths, and can also be used to find the initial values for the flow depth in various theoretical and numerical schemes.

Accurate prediction of pier scour can lead to economic design of bridge piers and prevent catastrophic incidents. This paper presents the application of self-adaptive evolutionary extreme learning machine (SAELM) to develop a new model for the prediction of local scour around bridge piers using 476 field pier scour measurements with four shapes of piers: sharp, round, cylindrical, and square. The model network parameters are optimized using the differential evolution algorithm. The best SAELM model calculates the scour depth as a function of pier dimensions and the sediment mean diameter. The developed SAELM model had the lowest error indicators when compared to regression-based prediction models for root mean square error (RMSE) (0.15, 0.65, respectively) and mean absolute relative error (MARE) (0.50, 2.0, respectively). The SAELM model was found to perform better than artificial neural networks or support vector machines on the same dataset. Parametric analysis showed that the new model predictions are influenced by pier dimensions and bed-sediment size and produce similar trends of variations of scour-hole depth as reported in literature and previous experimental measurements. The prediction uncertainty of the developed SAELM model is quantified and compared with existing regression-based models and found to be the least, ±0.03 compared with ±0.10 for other models.