Bridge pier live bed scour depth predictor with alert generation

Abstract

The present disclosure provides a system (100) and method (300) for predicting live bed scour depth around a bridge pier. The system (100) includes a portable device (102) having a processor (104), a memory (106) storing scour measurement data and scour depth relationship, a normalizer (116) for transforming parameters to normalized values, a parameter selector (120) for identifying parameter combination by minimizing gamma value and V_ratio value, a computation unit (122) for generating scour depth ratio (ds / y), an input interface (114) for receiving hydraulic and sediment parameters, a display interface (124) for presenting results, and an alert unit (126) for generating warning signal when the scour depth ratio (ds / y) exceeds predefined safety threshold. Unlike conventional empirical approaches with limited accuracy under field conditions, this approach enables accurate live bed scour depth prediction through physics-informed scour depth relationship, achieving reliable assessment for bridge pier safety applications.

Description

TECHNICAL FIELD

[0001] The present disclosure relates to the field of hydraulic engineering. Moreparticularly, the present disclosure relates to a bridge pier live bed scour depthpredictor with alert generation for assessing sediment erosion around bridgefoundations.BACKGROUND

[0002] Bridge pier scouring, where sediment erodes from around bridgefoundations due to complex interactions between flowing water and riverbedmaterials, poses significant threats to structural integrity of bridges and hashistorically contributed to catastrophic bridge failures resulting in substantialeconomic losses and endangerment of human lives, and accurate prediction ofscour depth under live bed scouring conditions where riverbed sediment remainsin active transport presents considerable challenges for field engineers andresearchers involved in bridge safety assessment and maintenance planning.

[0003] Conventional approaches for predicting scour depth around bridge piershave relied upon empirical relationships derived from laboratory experiments andlimited field observations, however such empirical formulations often exhibitinadequate accuracy when applied to diverse real-world field conditions due totheir inability to capture complex non-linear interactions among geometricparameters of pier structures, hydraulic flow characteristics, and sedimenttransport properties, and furthermore these conventional approaches typicallyrequire extensive computational resources or specialized expertise forimplementation, thereby limiting their practical utility for field engineersrequiring rapid and reliable scour depth assessments at bridge sites.

[0004] There exists a requirement for an improved approach that enables accurateprediction of live bed scour depth around bridge piers through effectiveintegration of hydraulic and sediment parameters while providing real-timeassessment capabilities and generating timely warnings when predicted scourdepths approach critical safety thresholds, thereby facilitating proactiveimplementation of countermeasures to safeguard bridge infrastructure.OBJECTS OF THE PRESENT DISCLOSURE

[0005] An object of the present disclosure is to provide accurate prediction of livebed scour depth around bridge piers through effective processing of hydraulic,geometric, and sediment parameters for reliable assessment of scouring conditionsat bridge sites.

[0006] Another object of the present disclosure is to provide timely alertgeneration when predicted scour depth exceeds predefined safety thresholds,thereby enabling proactive implementation of countermeasures to safeguardbridge pier foundations.

[0007] Yet another object of the present disclosure is to provide a portable andfield-deployable solution that facilitates rapid scour depth assessment withoutdependency on complex computational infrastructure or specialized expertise.SUMMARY

[0008] In an aspect, the present disclosure provides a bridge pier live bed scourdepth predictor with alert generation, including a portable device having aprocessor, a memory coupled to the processor storing scour measurement data andscour depth relationship, a normalizer coupled to the memory for transformingreceived parameters to normalized values, a parameter selector coupled to thenormalizer for identifying optimal parameter combination by minimizing gammavalue and V_ratio value, a computation unit coupled to the parameter selector andthe memory for applying the scour depth relationship to generate scour depthratio, an input interface coupled to the processor and the normalizer for receivinghydraulic and sediment parameters, a display interface coupled to the computationunit for presenting the scour depth ratio, and an alert unit coupled to thecomputation unit for generating warning signal when the scour depth ratioexceeds predefined safety threshold.

[0009] In another aspect, the present disclosure provides predicting live bed scourdepth around bridge pier using a portable device, including receiving hydraulicand sediment parameters, transforming each received parameter to normalizedvalues using minimum and maximum values from scour measurement data,identifying a parameter combination from the transformed parameters, applyingscour depth relationship to the parameter combination to generate scour depthratio, presenting the scour depth ratio, and generating warning signal when thescour depth ratio exceeds predefined safety threshold.BRIEF DESCRIPTION OF DRAWINGS

[0010] The accompanying drawings are included to provide a furtherunderstanding of the present disclosure and are incorporated in and constitute apart of this specification. The drawings illustrate exemplary embodiments of thepresent disclosure and, together with the description, serve to explain theprinciples of the present disclosure. The diagrams are for illustration only, whichthus is not a limitation of the present disclosure.

[0011] FIG. 1 illustrates a block diagram representation of a portable device forpredicting live bed scour depth around bridge pier with alert generation, inaccordance with an embodiment of the present disclosure.

[0012] FIG. 2 illustrates a phase diagram representation depicting development,deployment, and user operation phases for the portable device, in accordance withan embodiment of the present disclosure.

[0013] FIG. 3 illustrates a flowchart representation of a method for predicting livebed scour depth around bridge pier using the portable device, in accordance withan embodiment of the present disclosure.DETAILED DESCRIPTION

[0014] The ensuing description provides exemplary embodiments only, and is notintended to limit the scope, applicability, or configuration of the disclosure.Rather, the ensuing description of the exemplary embodiments will provide thoseskilled in the art with an enabling description for implementing an exemplaryembodiment. It should be understood that various changes may be made in thefunction and arrangement of elements without departing from the spirit and scopeof the disclosure as set forth.Definitions:Scour Depth Ratio: A dimensionless parameter represented as ds / y, where dscorresponds to scour depth and y corresponds to flow depth, that can quantifyextent of sediment erosion around a bridge pier relative to approach flowconditions, and can enable standardized comparison across different bridge sitesand flow scenariosLive Bed Scouring: A scouring condition that can occur when approach flowvelocity exceeds sediment incipient velocity (V / Vc > 1.0), where riverbedsediment remains in active transport while erosion occurs around bridge pierfoundations, and can be distinguished from clear water scouring conditions whereupstream sediment transport remains absent.Gamma Test: A statistical technique for determining optimal input parametercombinations in non-linear modeling applications by evaluating gamma value andV_ratio value, where minimum gamma value and V_ratio value can indicateparameter combinations that exhibit strongest correlation with output variable.Predefined Safety Threshold: A scour depth ratio (ds / y) value within a range of0.8 to 1.8, preferably 1.0, representing maximum acceptable scour depth ratio forsafe bridge pier operation, beyond which the alert unit generates warning signal,where the threshold value can be adjusted based on pier foundation depth,structural load capacity, and site-specific geological conditions.

[0015] An aspect of the present disclosure relates to a system for predicting livebed scour depth around a bridge pier that can include but may not be limited to aportable device including a processor, a memory coupled to the processor storingscour measurement data and scour depth relationship, a normalizer coupled to thememory for transforming received parameters to normalized values, a parameterselector coupled to the normalizer for identifying a parameter combination byminimizing gamma value and V_ratio value, a computation unit coupled to theparameter selector and the memory for applying the scour depth relationship togenerate scour depth ratio, an input interface coupled to the processor and thenormalizer for receiving hydraulic and sediment parameters, a display interfacecoupled to the computation unit for presenting the scour depth ratio, and an alertunit coupled to the computation unit for generating warning signal when the scourdepth ratio exceeds a predefined safety threshold.

[0016] Various embodiments of the present disclosure can be described usingFIGs. 1 to 3.

[0017] FIG. 1 illustrates a block diagram representation of a portable device forpredicting live bed scour depth around bridge pier with alert generation, inaccordance with an embodiment of the present disclosure.

[0018] Referring to FIG. 1, a system (100) for predicting live bed scour deptharound a bridge pier can include but may not be limited to a portable device (102)including a processor (104), a memory (106), a USB interface (108), anaccelerator (110), an auto-execution unit (112), an input interface (114), an inputvalidator (114-2), a normalizer (116), a sensor interface (118), a parameterselector (120), a computation unit (122), a display interface (124), an alert unit(126), and a threshold adjuster (126-2). The system (100) can establish a scourdepth prediction architecture by parameter normalization at the normalizer (116),parameter combination identification at the parameter selector (120), and scourdepth ratio computation at the computation unit (122) that can generate accuratepredictions for live bed scouring conditions while the alert unit (126) may providewarning signals when a predicted scour depth ratio exceeds a predefined safetythreshold.

[0019] In an embodiment, the system (100) can include the portable device (102)that can provide a self-contained prediction platform for assessing live bed scourdepth around bridge piers under field conditions. The portable device (102) canfunction by integrated processing capabilities through the processor (104) thatmay coordinate operations among the memory (106), the normalizer (116), theparameter selector (120), and the computation unit (122), through portable formfactor that can enable field deployment at bridge sites without requirement forexternal computational infrastructure, and to enable plug-and-play operation thatmay facilitate rapid scour depth assessment by field engineers and researchers.The portable device (102) can be configured as a USB-based device that mayinclude flash storage capacity within a range of 8 GB to 32 GB, preferably 16 GBfor storing the scour measurement data and the scour depth relationship within thememory (106), where the portable device (102) dimensions can accommodateneural accelerator hardware for executing computation operations whilemaintaining portability suitable for field deployment conditions.

[0020] In an embodiment, the portable device (102) can include the processor(104) that can coordinate operations among components of the portable device(102) for executing scour depth prediction operations including data flowmanagement, parameter processing coordination, and output direction through thedisplay interface (124) and the alert unit (126).

[0021] In an embodiment, the portable device (102) can include the memory(106) coupled to the processor (104), where the memory (106) can store scourmeasurement data and scour depth relationship correlating pier width, flow depth,and Froude number. The memory (106) can function by maintaining the scourmeasurement data compiled from experimental and field datasets that mayprovide reference values for parameter normalization operations, by storing thescour depth relationship that can correlate dimensional and dimensionlessparameters with scour depth ratio for prediction computations, and by providingminimum and maximum values from the scour measurement data that may enablethe normalizer (116) to transform received parameters to a normalized range. Thescour measurement data stored within the memory (106) can include datasetswithin a range of 200 to 400 datasets, preferably 300 datasets collected fromlaboratory experiments and field observations covering pier width to flow depthratio, approach flow velocity to sediment incipient velocity ratio, critical Froudenumber, pier width to median sediment size ratio, and geometric standarddeviation of bed material across ranges representative of live bed scouringconditions encountered at bridge sites.

[0022] In an embodiment, the memory (106) can further store a sedimentincipient velocity relationship correlating flow depth and median sediment size,and where the computation unit (122) can compute a sediment incipient velocityratio based on the sediment incipient velocity relationship. The sediment incipientvelocity relationship can function to enable computation of sediment incipientvelocity based on flow depth and median sediment size parameters that may beobtained from field measurements, to provide the sediment incipient velocityrelationship expressed as Vc = K x y^(1 / 6) x d50^(1 / 3) where K can represent acoefficient within a range of 5.5 to 7.0, preferably 6.36, where Vc corresponds tosediment incipient velocity, y corresponds to flow depth, and d50 corresponds tomedian sediment size, and to enable the computation unit (122) to derive theapproach flow velocity to sediment incipient velocity ratio when approach flowvelocity gets provided through the input interface (114) or measured through thesensor interface (118).

[0023] In an embodiment, the input interface (114) can receive the approach flowvelocity to sediment incipient velocity ratio (V / Vc) directly as a pre-computeddimensionless parameter, or alternatively can receive approach flow velocity (V),flow depth (y), and median sediment size (d50) as separate inputs, where thecomputation unit (122) computes sediment incipient velocity (Vc) using thesediment incipient velocity relationship and derives the approach flow velocity tosediment incipient velocity ratio (V / Vc) therefrom, thereby supporting both directparameter entry and computed parameter derivation modes of operation.

[0024] In an embodiment, the portable device (102) can include the USB interface(108) coupled to the processor (104) to establish connection with an externalcomputing device. The USB interface (108) can function to provide physical andelectrical connection capability that may enable the portable device (102) tointerface with laptop computers, desktop systems, or other computing devices atfield locations, to enable power supply to the portable device (102) through USBconnection that can eliminate requirement for a separate power source during fieldoperations, and to facilitate data communication between the portable device(102) and the external computing device that may enable display of predictionresults and user interaction through an external computing device interface.

[0025] In an embodiment, the portable device (102) can further include theaccelerator (110) coupled to the processor (104) and the computation unit (122),where the accelerator (110) can execute computation operations for generating thescour depth ratio. The accelerator (110) can function to provide dedicatedhardware resources for executing neural network inference operations that mayaccelerate scour depth ratio computation compared to general-purpose processorexecution, to enable real-time prediction capability that can support rapidassessment of multiple scenarios during field visits, and to offload computation-intensive operations from the processor (104) that may improve overall systemresponsiveness during user interactions.

[0026] In an embodiment, the portable device (102) can further include the auto-execution unit (112) coupled to the processor (104), where the auto-execution unit(112) can initiate the input interface (114) upon connection of the portable device(102) to the external computing device. The auto-execution unit (112) canfunction to detect USB connection establishment through the USB interface (108)that may trigger automatic initialization of prediction software, to launch agraphical user interface through the input interface (114) that can presentparameter entry fields to users without requirement for manual softwareinstallation or configuration, and to enable plug-and-play operation that mayreduce technical expertise required for conducting scour depth predictions at fieldsites.

[0027] In an embodiment, the portable device (102) can include the inputinterface (114) coupled to the processor (104) and the normalizer (116), where theinput interface (114) can receive a pier width to flow depth ratio (b / y), anapproach flow velocity to sediment incipient velocity ratio (V / Vc), a criticalFroude number (Frc), a pier width to median sediment size ratio (b / d50), and ageometric standard deviation of bed material (σg). The input interface (114) canfunction to present graphical interface elements including but not limited to entryfields for each of five dimensionless parameters that may guide users throughparameter entry process, to accept numeric values entered by users or receivedfrom external measurement systems that can provide parameter data for predictionoperations, and to direct received parameters to the normalizer (116) fortransformation prior to parameter combination identification at the parameterselector (120).

[0028] In an embodiment, the input interface (114) can operate in a direct inputmode where the approach flow velocity to sediment incipient velocity ratio(V / Vc) is received as a pre-computed dimensionless parameter, or in a computedinput mode where approach flow velocity (V), flow depth (y), and mediansediment size (d50) are received separately and the computation unit (122) derivesthe approach flow velocity to sediment incipient velocity ratio (V / Vc) using thesediment incipient velocity relationship stored in the memory (106).

[0029] In an embodiment, the input interface (114) can further include the inputvalidator (114-2) coupled to the normalizer (116), where the input validator (114-2) can verify each received parameter against minimum and maximum valuesstored in the memory (106) prior to transformation by the normalizer (116). Theinput validator (114-2) can function to compare each received parameter valueagainst corresponding minimum and maximum bounds maintained within thememory (106) that may establish valid parameter ranges for prediction operations,to generate an error indication when received parameter values fall outside validranges that can prompt users to verify and correct parameter entries, and to ensureonly validated parameters proceed to the normalizer (116) for transformation thatmay prevent erroneous predictions resulting from out-of-range input values.

[0030] In an embodiment, the portable device (102) can include the normalizer(116) coupled to the memory (106), where the normalizer (116) can transformeach received parameter to a value between a lower bound and an upper boundusing minimum and maximum values from the scour measurement data, wherethe lower bound can be within a range of 0.01 to 0.10, preferably 0.05, and theupper bound can be within a range of 0.90 to 0.99, preferably 0.95. Thenormalizer (116) can function to retrieve minimum and maximum values for eachparameter from the scour measurement data stored in the memory (106) that mayprovide reference bounds for transformation operations, to apply a normalizationexpression a_norm = lower bound + scaling factor x (a - a_min) / (a_max - a_min)to each received parameter, where the scaling factor can be 0.9, corresponding tothe difference between the upper bound of 0.95 and the lower bound of 0.05 thatcan map parameter values to a standardized range suitable for prediction modelinput, and to output transformed parameter values to the parameter selector (120)for identification of an optimal parameter combination. The normalization to therange between the lower bound and the upper bound can prevent boundary effectsthat may occur at extreme normalized values of 0 and 1, thereby improvingnumerical stability during computation operations performed by the computationunit (122).

[0031] In an embodiment, when received parameter values exceed the maximumvalues or fall below the minimum values from the scour measurement data storedin the memory (106), the normalization expression can produce transformedvalues outside the 0.05 to 0.95 range, where the physics-informed neural networkmodel implemented by the computation unit (122) can process such extrapolatednormalized values due to embedded physical constraints that maintain predictionvalidity for parameter combinations beyond the training data range.

[0032] In an embodiment, the portable device (102) can further include the sensorinterface (118) coupled to the processor (104), where the sensor interface (118)can receive flow measurements from an external sensor (128) positioned at abridge pier. The sensor interface (118) can function to establish communicationwith the external sensor (128) positioned at bridge pier locations that may providereal-time measurement data for prediction operations, to receive flow velocity andflow depth measurements from a flow sensor (128-2) of the external sensor (128)that can provide hydraulic parameters for scour depth prediction, and to receivesediment size and distribution measurements from a sediment sensor (128-4) ofthe external sensor (128) that may provide bed material characteristics forprediction computations.

[0033] In an embodiment, the portable device (102) can include the parameterselector (120) coupled to the normalizer (116), where the parameter selector (120)can identify a parameter combination from transformed parameters by minimizinga gamma value and a V_ratio value. The parameter selector (120) can function toevaluate transformed parameters received from the normalizer (116) using gammatest methodology that may assess suitability of parameter combinations for non-inear prediction modeling, to compute the gamma value and the V_ratio value forcandidate parameter combinations that can indicate correlation strength betweeninput parameters and scour depth ratio output, and to identify the parametercombination exhibiting minimum gamma value and minimum V_ratio value thatmay provide optimal configuration for scour depth prediction. The gamma testperformed by the parameter selector (120) can determine that the parametercombination including the pier width to flow depth ratio (b / y), the approach flowvelocity to sediment incipient velocity ratio (V / Vc), the critical Froude number(Frc), the pier width to median sediment size ratio (b / d50), and the geometricstandard deviation of bed material (σg) provides optimal configuration for livebed scour depth prediction based on minimum gamma value and minimumV_ratio value criteria.

[0034] In an embodiment, the portable device (102) can include the computationunit (122) coupled to the parameter selector (120) and the memory (106), wherethe computation unit (122) can receive the parameter combination, apply the scourdepth relationship to the parameter combination, and generate a scour depth ratio(ds / y). The computation unit (122) can function to receive the identified parametercombination from the parameter selector (120) that may include normalizedvalues for the pier width to flow depth ratio (b / y), the approach flow velocity tosediment incipient velocity ratio (V / Vc), the critical Froude number (Frc), the pierwidth to median sediment size ratio (b / d50), and the geometric standard deviationof bed material (σg), to retrieve the scour depth relationship from the memory(106) that can provide a prediction model correlating input parameters with scourdepth ratio output, and to execute prediction computation using the scour depthrelationship applied to the parameter combination that may generate a scour depthratio value representing predicted scour depth normalized by flow depth.

[0035] In an embodiment, the scour depth relationship stored in the memory (106)can implement a physics-informed neural network model that may combine data-driven learning with physical principles governing scour phenomena aroundbridge piers. The scour depth relationship can function to incorporate a simplifiedgoverning equation expressed as ds / y = C x (b / y)^α x Fr^β, where C can representa coefficient within a range of 1.5 to 2.5, preferably 2.0, where α can represent afirst exponent within a range of 0.55 to 0.75, preferably 0.65, and where β canrepresent a second exponent within a range of 0.35 to 0.55, preferably 0.43, thatmay provide physics-based constraint for prediction model training, to implementa total loss function combining a data loss component and a physics losscomponent that can ensure predictions satisfy both observed data patterns andphysical principles, and to provide prediction capability for the scour depth ratio(ds / y) within a range of 0.20 to 2.20, preferably 0.23 to 2.14 under live bedscouring conditions where approach flow velocity exceeds sediment incipientvelocity.

[0036] In an embodiment, the scour depth relationship stored in the memory (106)can implement the total loss function for training the physics-informed neuralnetwork model, where the total loss function can combine the data losscomponent and the physics loss component expressed as Total Loss = Loss_data +λ x Loss_physics, and where λ can represent a weighting factor within a range of0.05 to 0.20, preferably 0.1 that may balance importance of data-driven learningand physics-informed constraint during model training. The total loss function canfunction to ensure the computation unit (122) generates predictions satisfying bothobserved data patterns from the scour measurement data and physical principlesgoverning scour phenomena, to enable model convergence toward solutions thatmay accurately reproduce experimental observations while remaining consistentwith established hydraulic relationships, and to provide a training objective thatcan guide neural network parameter optimization toward physically plausiblescour depth ratio predictions.

[0037] In an embodiment, the data loss component of the total loss function canbe expressed as Loss_data = (1 / N) x Σ(ds_predicted - ds_observed)2 where N canrepresent a number of data points from the scour measurement data stored in thememory (106), ds_predicted can represent scour depth predicted by the physics-informed neural network model, and ds_observed can represent observed scourdepth from experimental and field datasets. The data loss component can functionto quantify mean squared error between predicted and observed scour depthvalues that may drive the physics-informed neural network model toward accuratereproduction of measured data, to provide gradient information during trainingthat can guide parameter adjustment toward minimizing prediction errors, and toensure the scour depth relationship learns patterns present within the scourmeasurement data for generating accurate scour depth ratio predictions throughthe computation unit (122).

[0038] In an embodiment, the physics loss component of the total loss functioncan be expressed as Loss_physics = (1 / M) x Σ(ds_NN - ds_HEC-18)2 where Mcan represent a number of collocation points, ds_NN can represent scour depthpredicted by the physics-informed neural network model at the collocation points,and ds_HEC-18 can represent scour depth computed from the simplifiedgoverning equation ds / y = C x (b / y)^α x Fr^β at corresponding collocation points.The physics loss component can function to enforce physical constraint at thecollocation points distributed across parameter space that may ensure neuralnetwork predictions comply with the established scour depth relationship, topenalize deviations between neural network output and physics-based predictionthat can guide model toward physically consistent solutions, and to enableaccurate extrapolation beyond training data range by embedding governingphysics within the scour depth relationship stored in the memory (106).

[0039] In an embodiment, the computation unit (122) can compute an errorbetween predicted scour depth ratio and observed scour depth ratio stored in thememory (106), and compute a deviation between the predicted scour depth ratioand the scour depth relationship. The computation unit (122) can function tocompare the predicted scour depth ratio against observed values from fielddatasets stored in the memory (106) that may quantify prediction accuracythrough error metrics including but not limited to mean absolute percentage error,root mean square error, and coefficient of determination values that cancharacterize prediction performance, and to evaluate the deviation betweenpredicted values and physics-based scour depth relationship that may assessphysical consistency of predictions. The coefficient of determination can bewithin a range of 0.92 to 0.98, preferably 0.9251, and the root mean square errorcan be within a range of 0.08 to 0.15, preferably 0.1116 for the physics-informedneural network model predictions.

[0040] In an embodiment, the portable device (102) can include the displayinterface (124) coupled to the computation unit (122), where the display interface(124) can present the scour depth ratio (ds / y). The display interface (124) canfunction to receive the computed scour depth ratio from the computation unit(122) that may represent predicted scour depth normalized by flow depth forassessed bridge pier conditions, to render graphical presentation of predictionresults that can include the scour depth ratio value and associated error metrics,and to provide visual indication of prediction confidence that may assist fieldengineers in assessing reliability of scour depth estimates for decision-makingpurposes.

[0041] In an embodiment, the portable device (102) can include the alert unit(126) coupled to the computation unit (122), where the alert unit (126) cangenerate a warning signal when the scour depth ratio (ds / y) exceeds thepredefined safety threshold. The alert unit (126) can function to receive thecomputed scour depth ratio from the computation unit (122) that may becompared against the predefined safety threshold maintained within the portabledevice (102), to evaluate whether the computed scour depth ratio exceeds thepredefined safety threshold that can indicate elevated scouring risk requiringattention, and to generate the warning signal through visual, audible, or electronicnotification that may alert field personnel to potential safety concerns at anassessed bridge pier location.

[0042] In an embodiment, the alert unit (126) can further include the thresholdadjuster (126-2) coupled to the computation unit (122), where the thresholdadjuster (126-2) can permit modification of the predefined safety threshold basedon pier geometry and flow velocity at the bridge pier. The threshold adjuster (126-2) can function to provide an interface for adjusting the predefined safetythreshold value that may accommodate site-specific conditions at different bridgelocations, to enable threshold modification based on pier geometry characteristicsincluding but not limited to pier diameter and foundation depth that can influenceacceptable scour depth limits, and to permit threshold adjustment based on flowvelocity conditions that may vary seasonally or during flood events at bridge sites.

[0043] In an embodiment, the system (100) can include the external sensor (128)coupled to the sensor interface (118) of the portable device (102), where theexternal sensor (128) can provide field measurement capability for obtaininghydraulic and sediment parameters at bridge pier locations. The external sensor(128) can include the flow sensor (128-2) that can measure approach flow velocityand flow depth at a bridge pier location, and the sediment sensor (128-4) that canmeasure median sediment size and geometric standard deviation of bed material.The flow sensor (128-2) can be configured as an acoustic doppler velocimeter orsimilar flow measurement device that may provide real-time hydraulic parameterdata to the sensor interface (118). The sediment sensor (128-4) can be configuredas an acoustic backscatter sensor or similar sediment characterization device thatmay provide bed material properties to the sensor interface (118) for scour depthprediction operations.

[0044] FIG. 2 illustrates a phase diagram representation depicting development,deployment, and user operation phases for the portable device, in accordance withan embodiment of the present disclosure.

[0045] Referring to FIG. 2, a phase diagram (200) can demonstrate operationalphases for the system (100) that may include a Phase I (202) corresponding todevelopment and deployment operations and a Phase II (212) corresponding touser operation activities. The phase diagram (200) can illustrate progression frominitial development through field deployment that may enable plug-and-play scourdepth prediction capability at bridge sites.

[0046] In an embodiment, the Phase I (202) can include a design interface block(204) that may involve creating the graphical user interface for the input interface(114), where the design interface block (204) can include entry fields for the pierwidth to flow depth ratio (b / y), the approach flow velocity to sediment incipientvelocity ratio (V / Vc), the critical Froude number (Frc), the pier width to mediansediment size ratio (b / d50), and the geometric standard deviation of bed material(σg) along with prediction initiation controls. The Phase I (202) can furtherinclude an integrate components block (206) that may involve linking the inputinterface (114) with the normalizer (116), the parameter selector (120), thecomputation unit (122), and the scour depth relationship stored in the memory(106), where the integrate components block (206) can ensure seamless data flowfrom parameter entry through prediction computation to result presentation.

[0047] In an embodiment, the Phase I (202) can include a deploy on portabledevice block (208) that may involve loading integrated software and the scourdepth relationship onto the portable device (102) including the memory (106) withthe scour measurement data and prediction model files, where the deploy onportable device block (208) can configure the auto-execution unit (112) forautomatic interface launch upon USB connection. The Phase I (202) can furtherinclude a test and validate ready block (210) that may involve verifying predictionaccuracy using sample parameter inputs and confirming plug-and-playfunctionality through insertion into the external computing device, where the testand validate ready block (210) can ensure the portable device (102) generatesaccurate scour depth ratio predictions matching expected values from validationdatasets.

[0048] In an embodiment, the Phase II (212) can include an insert portable deviceblock (214) where a user connects the portable device (102) to the externalcomputing device through the USB interface (108), an auto launch interface block(216) where the auto-execution unit (112) detects USB connection and initiatesthe input interface (114), an enter parameters block (218) where the user entersthe five dimensionless parameters through the input interface (114), a validateinputs block (220) where the input validator (114-2) verifies each parameteragainst minimum and maximum values stored in the memory (106), a processthrough portable device block (222) where the normalizer (116) transformsparameters, the parameter selector (120) identifies optimal parametercombination, and the computation unit (122) generates the scour depth ratio(ds / y), and a generate outputs block (224) where the computation unit (122)produces the scour depth ratio prediction along with associated error metrics.

[0049] In an embodiment, the Phase II (212) can include a display results block(226) that may involve the display interface (124) presenting the computed scourdepth ratio (ds / y) and the error metrics to the user through the graphical userinterface rendered on the external computing device display. The Phase II (212)can further include a user review block (228) where the user can evaluateprediction results and the warning signal generated by the alert unit (126) whenthe scour depth ratio (ds / y) exceeds the predefined safety threshold, enablinginformed decision-making regarding countermeasure implementation at theassessed bridge pier location.

[0050] In an embodiment, the Phase II (212) can include an eject portable deviceblock (230) that may involve the user safely disconnecting the portable device(102) from the external computing device through the USB interface (108)following completion of scour depth prediction operations, where the portabledevice (102) can be transported to subsequent bridge sites for additionalassessments or stored for future use.

[0051] FIG. 3 illustrates a flowchart representation of a method for predicting livebed scour depth around bridge pier using the portable device, in accordance withan embodiment of the present disclosure.

[0052] Referring to FIG. 3, a method (300) for predicting live bed scour deptharound bridge pier using the portable device (102) can include but may not belimited to a block (302), receiving the pier width to flow depth ratio (b / y), theapproach flow velocity to sediment incipient velocity ratio (V / Vc), the criticalFroude number (Frc), the pier width to median sediment size ratio (b / d50), and thegeometric standard deviation of bed material (σg), a block (304), transformingeach received parameter to a value between the lower bound and the upper boundusing minimum and maximum values from the scour measurement data stored inthe memory (106), a block (306), identifying the parameter combination fromtransformed parameters by minimizing the gamma value and the V_ratio value, ablock (308), applying the scour depth relationship stored in the memory (106) tothe parameter combination to generate the scour depth ratio (ds / y), a block (310),presenting the scour depth ratio (ds / y) on the display interface (124), and a block(312), generating the warning signal when the scour depth ratio (ds / y) exceeds thepredefined safety threshold.

[0053] In an embodiment, the block (302), receiving the pier width to flow depthratio (b / y), the approach flow velocity to sediment incipient velocity ratio (V / Vc),the critical Froude number (Frc), the pier width to median sediment size ratio(b / d50), and the geometric standard deviation of bed material (σg) can function toaccept parameter values entered through the input interface (114) by users whomay obtain values from field measurements or design specifications, to receiveparameter values from the sensor interface (118) when the external sensor (128)including the flow sensor (128-2) and the sediment sensor (128-4) get deployed ata bridge pier location, and to validate received parameter values through the inputvalidator (114-2) that can verify values fall within acceptable ranges defined bythe minimum and maximum values stored in the memory (106).

[0054] In an embodiment, the block (304), transforming each received parameterto a value between the lower bound and the upper bound using minimum andmaximum values from the scour measurement data stored in the memory (106)can function to retrieve the minimum and maximum values for the pier width toflow depth ratio (b / y), the approach flow velocity to sediment incipient velocityratio (V / Vc), the critical Froude number (Frc), the pier width to median sedimentsize ratio (b / d50), and the geometric standard deviation of bed material (σg) fromthe scour measurement data, to apply the normalization expression a_norm =lower bound + scaling factor x (a - a_min) / (a_max - a_min) to each receivedparameter where a represents received parameter value, a_min representsminimum value, and a_max represents maximum value from the scourmeasurement data, and to output transformed parameter values that may fallwithin or outside the range of the lower bound to the upper bound depending onwhether received parameter values fall within or exceed the minimum andmaximum values from the scour measurement data, that may provide standardizedinputs for parameter combination identification and scour depth ratiocomputation.

[0055] In an embodiment, the block (304) can be performed by the normalizer(116) coupled to the memory (106), where the normalizer (116) can access theminimum and maximum values maintained within the scour measurement data foreach of the five dimensionless parameters. The normalization range of the lowerbound to the upper bound can prevent boundary effects that may occur at extremenormalized values of 0 and 1, thereby improving numerical stability duringcomputation operations performed by the computation unit (122).

[0056] In an embodiment, the block (306), identifying the parameter combinationfrom the transformed parameters by minimizing the gamma value and the V_ratiovalue can function to evaluate the transformed parameter values using gamma testmethodology that may assess near-neighbor relationships between inputparameters and scour depth ratio output, to compute the gamma valuerepresenting noise variance estimate and the V_ratio value representing ratio ofgamma to output variance that can indicate parameter combination suitability fornon-linear prediction, and to select the parameter combination exhibitingminimum gamma value and minimum V_ratio value that may provide optimalconfiguration for accurate scour depth ratio prediction.

[0057] In an embodiment, the block (306) can be performed by the parameterselector (120) coupled to the normalizer (116), where the parameter selector (120)can evaluate candidate parameter combinations to determine configurationproviding strongest correlation with the scour depth ratio (ds / y). The gamma testperformed at the block (306) can determine that all five parameters including thepier width to flow depth ratio (b / y), the approach flow velocity to sedimentincipient velocity ratio (V / Vc), the critical Froude number (Frc), the pier width tomedian sediment size ratio (b / d50), and the geometric standard deviation of bedmaterial (σg) contribute to the optimal parameter combination for live bed scourdepth prediction.

[0058] In an embodiment, the block (308), applying the scour depth relationshipstored in the memory (106) to the parameter combination to generate the scourdepth ratio (ds / y) can function to retrieve the scour depth relationship from thememory (106) that may implement the physics-informed neural network modelcorrelating input parameters with scour depth ratio output, to process theidentified parameter combination through the scour depth relationship that canperform inference computation using trained model weights and activationfunctions, and to generate the scour depth ratio value representing predicted scourdepth normalized by flow depth for assessed bridge pier conditions under live bedscouring.

[0059] In an embodiment, the block (308) can be performed by the computationunit (122) coupled to the parameter selector (120) and the memory (106), wherethe computation unit (122) can execute prediction computation using theaccelerator (110) for improved processing speed. The scour depth relationshipapplied at the block (308) can implement physics-informed approach combiningdata-driven learning from the scour measurement data with physical constraintbased on the simplified governing equation ds / y = C x (b / y)^α x Fr^β that mayensure predictions remain physically consistent with established scour phenomenaunderstanding.

[0060] In an embodiment, the block (308) can generate the scour depth ratio(ds / y) within the range of 0.20 to 2.20, preferably 0.23 to 2.14 for live bedscouring conditions where the approach flow velocity to sediment incipientvelocity ratio (V / Vc) exceeds 1.0, representing predicted scour depth valuesranging from shallow erosion to deep scour hole formation around bridge pierfoundations depending on hydraulic and sediment parameter combinations.

[0061] In an embodiment, the block (310), presenting the scour depth ratio (ds / y)on the display interface (124) can function to receive the computed scour depthratio from the computation unit (122) that may represent prediction result forassessed bridge pier conditions, to render graphical presentation through thedisplay interface (124) that can include the scour depth ratio value displayed innumeric format along with associated error metrics including but not limited tomean absolute percentage error, root mean square error, and coefficient ofdetermination, and to provide visual feedback to users that may enable assessmentof prediction results and confidence levels for decision-making purposes.

[0062] In an embodiment, the block (312), generating the warning signal whenthe scour depth ratio (ds / y) exceeds the predefined safety threshold can function tocompare the computed scour depth ratio against the predefined safety thresholdmaintained within the portable device (102) that may represent maximumacceptable scour depth ratio for safe bridge operation, to evaluate whether thecomputed scour depth ratio exceeds the predefined safety threshold indicatingelevated scouring risk at the assessed bridge pier location, and to produce thewarning signal through the alert unit (126) that can notify field personnel ofpotential safety concerns requiring attention or countermeasure implementation.

[0063] In an embodiment, the warning signal generated at the block (312) caninclude visual indication through the display interface (124) that may highlight thescour depth ratio value exceeding the predefined safety threshold, audible alertthrough external computing device audio output that can attract immediateattention of field personnel, and electronic notification that may be transmitted toremote monitoring systems for centralized bridge infrastructure management.

[0064] In an embodiment, the collocation points can represent specific locationswithin input parameter space where the physics constraint may get enforcedduring training of the scour depth relationship, where the collocation points can bedistributed across ranges of the pier width to flow depth ratio (b / y), the approachflow velocity to sediment incipient velocity ratio (V / Vc), the critical Froudenumber (Frc), the pier width to median sediment size ratio (b / d50), and thegeometric standard deviation of bed material (σg) that may correspond to live bedscouring conditions. The collocation points can function to provide evaluationlocations where neural network predictions can be compared against thesimplified governing equation output for computing the physics loss component,to enable physics-informed regularization that may prevent overfitting to trainingdata while maintaining physical consistency, and to ensure the computation unit(122) generates scour depth ratio predictions remaining within physicallyplausible bounds even for parameter combinations not present in the scourmeasurement data.

[0065] The described system (100) and method (300) can present an architecturethat may enable accurate prediction of live bed scour depth around bridge piersthrough integrated parameter normalization, optimal parameter combinationidentification, and physics-informed computation while providing real-timewarning capability when predicted scour depths approach critical safetythresholds. The portable device (102) can provide field-deployable predictioncapability that may assist engineers and researchers in assessing local scouringrisks at bridge sites and implementing timely precautions to minimize scour-related structural concerns.

[0066] While considerable emphasis has been placed herein on the preferredembodiments, it will be appreciated that many embodiments can be made and thatmany changes can be made in the preferred embodiments without departing fromthe principles of the disclosure. These and other changes in the preferredembodiments of the disclosure will be apparent to those skilled in the art from thedisclosure herein, whereby it is to be distinctly understood that the foregoingdescriptive matter is to be interpreted merely as illustrative of the disclosure andnot as a limitation.EXAMPLESExample 1: Scour depth ratio prediction accuracy assessment

[0067] The computation unit (122) of the portable device (102) applying the scourdepth relationship stored in the memory (106) can be evaluated for predictionaccuracy using field validation datasets from river bridge pier locations. Table 1presents error assessment metrics including mean absolute percentage error(MAPE), root mean square error (RMSE), and coefficient of determination (R2)for the scour depth ratio (ds / y) predictions.Table 1: Error assessment in live bed scour depth ratio (ds / y) calculationusing the scour depth relationship

[0068] The results demonstrate that the scour depth relationship implemented bythe computation unit (122) can achieve coefficient of determination (R2) valuesexceeding 0.92 for both field datasets, indicating strong correlation betweenpredicted and observed scour depth ratio (ds / y) values. The mean absolutepercentage error (MAPE) values below 13% and root mean square error (RMSE)values below 0.13 can confirm reliable prediction capability of the portable device(102) for field applications at bridge pier locations under live bed scouringconditions.Example 2: Sample calculation for bridge pier scour depth ratio prediction

[0069] The method (300) for predicting live bed scour depth around a bridge pierusing the portable device (102) can be demonstrated through a sample calculationfor a bridge pier located at Pearl River. Table 2 presents input parameters receivedat the block (302) and computed dimensionless parameters for the scour depthratio (ds / y) prediction.Table 2: Input parameters and computed dimensionless values for PearlRiver bridge pierObserved scour depth ratio (ds / y) 0.3636 3.0 / 8.25

[0070] The results demonstrate that the block (302) can receive the fivedimensionless parameters including the pier width to flow depth ratio (b / y) of0.3030, the approach flow velocity to sediment incipient velocity ratio (V / Vc) of1.053, the critical Froude number (Frc) of 0.4660, the pier width to mediansediment size ratio (b / d50) of 23.584, and the geometric standard deviation of bedmaterial (σg) of 2.898. The approach flow velocity to sediment incipient velocityratio (V / Vc) exceeding 1.0 can confirm live bed scouring condition at the bridgepier location.Example 3: Normalization and scour depth ratio prediction computation

[0071] The block (304) performed by the normalizer (116) can transform eachreceived parameter to a normalized value using the minimum and maximumvalues from the scour measurement data stored in the memory (106). Forparameters where received values exceed the maximum values in the scourmeasurement data, the normalization expression produces transformed valuesgreater than 0.95, demonstrating extrapolation capability of the physics-informedscour depth relationship. Table 3 presents normalized parameter values and thepredicted scour depth ratio (ds / y) generated at the block (308) by the computationunit (122)Table 3: Normalized parameters and predicted scour depth ratio for PearlRiver bridge pier

[0072] The results demonstrate that the normalizer (116) at the block (304) cantransform each received parameter to a normalized value within the range of thelower bound to the upper bound using the normalization expression a_norm =lower bound + scaling factor x (a - a_min) / (a_max - a_min). The normalizedvalues for geometric standard deviation (σg) of 1.38 and approach flow velocity tosediment incipient velocity ratio (V / Vc) of 1.2233 exceed the upper bound of0.95, indicating that the Pearl River bridge pier parameters for these variablesexceed the maximum values in the scour measurement data, where the physics-informed neural network model maintains prediction accuracy for suchextrapolated conditions as demonstrated by the close agreement between predictedand observed scour depth ratio values.ADVANTAGES OF THE PRESENT DISCLOSURE

[0073] The present disclosure provides the system and the method that can enableaccurate prediction of live bed scour depth ratio (ds / y) around bridge piersthrough the physics-informed scour depth relationship, thereby facilitating timelyassessment of scouring risks at bridge sites without requirement for complexcomputational infrastructure.

Claims

1. A system (100) for predicting live bed scour depth around bridge pier, comprising: a portable device (102) comprising: a processor (104); a memory (106) coupled to the processor (104), wherein the memory (106) stores scour measurement data and scour depth relationship correlating pier width, flow depth, and Froude number; a normalizer (116) coupled to the memory (106), wherein the normalizer (116) transforms each received parameter to a value between 0.05 and 0.95 using minimum and maximum values from the scour measurement data; a parameter selector (120) coupled to the normalizer (116), wherein the parameter selector (120) identifies a parameter combination from the transformed parameters by minimizing gamma value and V_ratio value; and a computation unit (122) coupled to the parameter selector (120) and the memory (106), wherein the computation unit (122) receives the parameter combination, applies the scour depth relationship to the parameter combination, and generates scour depth ratio (ds / y); an input interface (114) coupled to the processor (104) and the normalizer (116), wherein the input interface (114) receives pier width to flow depth ratio (b / y), approach flow velocity to sediment incipient velocity ratio (V / Vc), critical Froude number (Frc), pier width to median sediment size ratio (b / d50), and geometric standard deviation of bed material (σg); and a display interface (124) coupled to the computation unit (122), wherein the display interface (124) presents the scour depth ratio; and an alert unit (126) coupled to the computation unit (122), wherein the alert unit (126) generates warning signal when the scour depth ratio exceeds predefined safety threshold.

2. The system (100) as claimed in claim 1, wherein the portable device (102) comprises a USB interface (108) coupled to the processor (104) to establish connection with external computing device.

3. The system (100) as claimed in claim 2, wherein the portable device (102) further comprises an auto-execution unit (112) coupled to the processor (104), wherein the auto-execution unit (112) initiates the input interface (114) upon connection of the portable device (102) to the external computing device.

4. The system (100) as claimed in claim 1, wherein the portable device (102) further comprises an accelerator (110) coupled to the processor (104) and the computation unit (122), wherein the accelerator (110) executes computation operations for generating the scour depth ratio.

5. The system (100) as claimed in claim 1, wherein the portable device (102) further comprises a sensor interface (118) coupled to the processor (104), wherein the sensor interface (118) receives flow measurements from flow sensor (128-2) and sediment measurements from sediment sensor (128-4) positioned at the bridge pier.

6. The system (100) as claimed in claim 1, wherein the memory (106) further stores sediment incipient velocity relationship correlating flow depth and median sediment size, and wherein the computation unit (122) computes the sediment incipient velocity ratio (V / Vc) based on the sediment incipient velocity relationship.

7. The system (100) as claimed in claim 1, wherein the computation unit (122) computes error between predicted scour depth ratio and observed scour depth ratio stored in the memory (106), and computes deviation between the predicted scour depth ratio and the scour depth relationship.

8. The system (100) as claimed in claim 1, wherein the input interface (114) further comprises an input validator (114-2) coupled to the normalizer (116), wherein the input validator (114-2) verifies each received parameter against minimum and maximum values stored in the memory (106) prior to transformation by the normalizer (116).

9. The system (100) as claimed in claim 1, wherein the alert unit (126) further comprises a threshold adjuster (126-2) coupled to the computation unit (122), wherein the threshold adjuster (126-2) permits modification of the predefined safety threshold based on pier geometry and flow velocity at the bridge pier.

10. A method (300) for predicting live bed scour depth around bridge pier using a portable device (102), the method comprising: receiving (302) pier width to flow depth ratio (b / y), approach flow velocity to sediment incipient velocity ratio (V / Vc), critical Froude number (Frc), pier width to median sediment size ratio (b / d50), and geometric standard deviation of bed material (σg); transforming (304) each received parameter to a value between 0.05 and 0.95 using minimum and maximum values from scour measurement data stored in a memory (106); identifying (306) a parameter combination from the transformed parameters by minimizing gamma value and V_ratio value; applying (308) scour depth relationship stored in the memory (106) to the parameter combination to generate scour depth ratio (ds / y); presenting (310) the scour depth ratio on a display interface (124); and generating (312) warning signal when the scour depth ratio exceeds predefined safety threshold.