Development of a Safety-Inspection Method for River Bridges in Turkey

: In this paper, ﬁndings of a research project about river bridges in Turkey are shared and details of the developed safety-inspection method based on hydrological and hydraulic factors are presented. In the project, the Western Black Sea Basin was chosen as the pilot area, where the basin is mountainous with steep slopes and has a rainy climate with frequent ﬂash ﬂoods. Many river bridges in the basin were inspected at di ﬀ erent ﬂow conditions throughout the project duration of three years. The developed safety-inspection method is composed of four main parts: evaluation of watershed hydrology and its ﬂood potential, stream stability, bridge characteristics and a rapid scour assessment. A structural assessment is also included in the method. Five river bridges in the area were chosen for detailed inspection and application of the method. Results showed that the method was capable of identifying and ranking the bridges in regard to maintenance needs and forming a comprehensive inventory for bridge engineers.


Introduction
After failure of Caycuma Bridge pier in April 2012 that resulted in the death of 15 people, problems of river bridges attracted further attention in Turkey. Since then, severe damages or failures of some river bridges have been reported, especially in the Black Sea Basin of the country. This basin is mountainous with steep slopes and has a rainy climate with frequent flash floods. With an increasing number of bridge disasters, the General Directorate of Highways (GDH) introduced a bridge management system (BMS) in 2017 in which all structural data of bridges in Turkey were transferred to the system, with the information about the requirements of the inspection/repair needs of the bridges, and the cost of maintenance. In this system, each subsidiary component of the bridge is inspected and graded visually or with equipment, where necessary, according to deterioration, aging, damage and loss of elements. Meanwhile, a collaborative research project with GDH and two universities was conducted with duration of three years to investigate the problems of river bridges in Turkey with extensive field observations and measurements in the Western Black Sea Region [1]. The study noted that almost all bridge failures in Turkey are river bridges and these were caused by various hydraulic factors, mainly excessive scouring around bridge piers, and debris accumulation. Along with this study, Akay [2] developed a safety-inspection method for river bridges that could be implemented into the current BMS to inspect bridges, especially with regard to hydraulic effects, and to form an inventory consisting of hydrologic and hydraulic factors and their effects on the structure.
Many countries developed their own BMSs depending on their needs, to manage their bridge stock efficiently, while some countries adopted widely used well-known systems of other countries [3][4][5][6]. In a traditional BMS, there are four modules: inventory, inspection, maintenance and optimization. Among those modules, inventory is considered as the most important part in which detailed database of all attributes of the bridge elements are collected [7]. Thus, various countries all over the world, such as France, Italy, Taiwan and India are still attempting to develop their national bridge database to improve their asset management [8][9][10]. Currently consisting of 28-member countries, the European Union has a wide variety of infrastructure assets with different construction types and BMSs. The majority of those BMSs focused mainly on structural health monitoring with less consideration for water-related hazards, such as scour at bridge foundation, and debris accumulation [11]. Meanwhile, some studies were conducted on the extreme weather events and changes in the frequency of extreme flooding due to global warming and the potential risks they pose to bridges [12][13][14]. Thus, today emphasis is placed on the bridge scour assessment with the increase in the number of catastrophic failures of bridges associated with flood induced events. Various studies were conducted to assess stream stability and channel conditions, while, especially in the last decade or so, research has focused on the development of bridge scour assessment methods and the estimation of bridge scour vulnerability [15][16][17][18][19][20][21][22][23].
In 1995, Lagasse et al. [24] proposed a three-level approach to assess the channel stability in vicinity of bridges, in which geomorphic concepts and qualitative analysis in Level 1, hydrologic, hydraulic and sediment transport aspects in Level 2, and mathematical and physical modelling studies in Level 3 were assessed. This approach is implemented in HEC-20 which is a widely known manual for identifying stream stability problems and presenting guidelines for bridge owners in the USA. According to this approach, stream characteristics and land use changes are considered for an overall stability with the analysis of vertical and lateral stability and stream response, with a number of parameters considered for characterization of the watershed response and morphological evolution. Rosgen [25] proposed a method based on assessing the stability of a stable reach as a reference and then performing a departure analysis to an unstable reach of the same stream type. Montgomery and MacDonald [26] developed a diagnostic approach in which the system and its elements were defined and assessed. Johnson [27] developed a preliminary assessment method based on the previous research related with stream stability and rating channel stability in the vicinity of the bridges located across the USA. In that study, Johnson [27] assessed the channel stability using thirteen indicators of watershed and floodplain activity and characteristics, flow habit, channel pattern and entrenchment or channel confinement, following the concept of Montgomery and MacDonald [26] who stated that geomorphic conditions, such as confinement and entrenchment, played an important role in characterizing channel stability. Furthermore, Shields et al. [28] stated that watershed characteristics were crucial for assessing the existing and future channel stability. In all these approaches, watershed response was very limitedly considered, and focus was mainly on the stream stability. Moreover, various prediction methodologies, such as simulating the temporal variations of the total scour depth under unsteady flow conditions or introducing a dimensionless effective work done by the flow on the sediment bed around the pier, were developed for bridge scour assessment, which are applicable under complex hydraulic/hydrological conditions, and limited scour field data [29][30][31][32][33][34][35]. In a study, Bekic et al. [36] applied two different methods of bridge scour assessment, namely the Colorado method [37] and the modified BA 74/06 method [38], on 100 railway bridges located over watercourse and showed that the resulting scour rankings differed for 20% of bridges with a significant difference of ranking in 10% of cases.
Within the European framework programs, many projects were carried out to meet the demands of a growing bridge stock and the need for future traffic demands [39]. One of these projects was the BRIDGE SMS: Intelligent bridge maintenance and management system which aimed to provide standardized methods for the assessment of bridge scour risk and implement common procedures among different organizations [40]. In this method, standardized procedures for structural and scour inspection were developed, allowing for four different inspection levels for scour inspection and evaluation, with the complexity and comprehensiveness of the inspections increasing with the levels. Level 1-general inspection, is comprised of 11 structural elements and 11 scour components. In a case where a structure Level 1 inspection is not adequate, then the assessment is progressed to detailed inspection, in which 14 individual structural elements are considered throughout scour inspection in Level 2. This level evaluates and records general scour, constriction scour and local scour. The procedure for inspecting bridge elements and components include a detailed visual assessment [41] and a detailed scour inspection, requiring inputs of river bed levels, foundation depth and data on general river channel instabilities. The condition rating between very good (0) to failed (5) is calculated separately for the structural and scour aspect and is accompanied with the time for the next inspection and the assessment of maintenance/repair costs. The scour condition is graded by a custom algorithm in Level 1 and by a risk-based matrix in Level 2, combining the result of the general, constriction and total scour. In the literature, there are many methods and systems, ranging from simple to advanced levels, using sophisticated measuring equipment and techniques. Hence, each country has to decide on the scope of their BMS according to their needs, future projections and budget limits.
In Turkey, a practical, simple but comprehensive method that involves various parameters and characteristics of watershed to enable assessment of the hydrological behavior and flash flood potential of the watershed is necessary for successful asset management. Thus, a simple approach similar to that of Johnson [27] was adopted. However, many indicators of not just stream, but watershed, bridge and environmental characteristics were taken into consideration to form an inclusive database of the bridge stock, as done in more comprehensive systems. The method allowed for two levels, as in many systems. Level 1 is comprised of a basic assessment of all bridge stock while Level 2 is performed in case the result of the Level 1 inspection is found to be insufficient. The method lacks a detailed scour assessment method in Level 1, which is considered in Level 2.
The goals of this study are to share findings of the aforementioned collaborative research project and present the developed safety-inspection algorithm, especially based on hydraulic indicators for river bridges in Turkey. Many river bridges located in the Western Black Sea Basin were inspected at different flow conditions throughout the project duration of three years. According to the problems encountered and observations made, the developed safety-inspection method was composed of four main parts: evaluation of watershed hydrology and its flood potential, stream stability, bridge characteristics and a rapid scour assessment. A structural assessment algorithm different from the current method in the BMS was also incorporated. These comprise the Level 1 of the developed safety-inspection algorithm. Five river bridges whose inventory data are partially available and either had a serious maintenance work during their service life or still have some problems and were inspected from time to time by GDH were chosen in the pilot area. These bridges were inspected in detail throughout the project duration to test the applicability of the method and its capability.

Field Study
Many river bridges in the Western Black Sea Basin were observed at different flow conditions throughout the study, about ten times in three years. During field trips, characteristics of the watersheds, stream flow, environmental or human induced factors affecting bridges and bridge characteristics were observed. Among those inspected bridges, five of them whose details are given in Table 1 and shown in Figure 1, were selected for full inspection and the application of the developed method.
The bathymetry of stream bed and scour holes around piers and abutments was also measured to track changes in stream beds in the vicinity of bridges, and possible settlements at the structures. Field measurements were conducted by a private company. Measurements of stream bed elevation including scour holes, coordinates of the location and mean water level were performed across and along the stream in the vicinity of the bridge concerned, at frequent intervals using a GPS coupled instrument situated at the reach of the stream. Similar measurements were also performed at the same coordinates during subsequent field visits. However, no noteworthy change in stream bed was noted during this three-year period. out that overall evaluation of river bridges should include stability indicators, starting at watershed characteristics or vertical and lateral aspects of the channel; the current method involved four main parts, all concentrated on hydrological and hydraulic factors. The first part involved a hydrological assessment of the watershed and flood potential while the second and third parts consisted of the stability indicators of stream and bridge characteristics, respectively, that were determined or modified using past studies in the literature, lessons learned from failures of river bridges in Turkey, and observations from field studies. During evaluation of the indicators, characteristics of the indicator in the upstream and downstream parts of the bridge were assessed. In the last part, a rapid scour check was carried out to determine scour susceptibility of the bridges. Besides hydrological and hydraulic assessment, a structural evaluation of the river bridges was also included, which was composed of evaluation of main body, and earth retaining and serviceability components of bridges in regard to loss, damage, aging and deformation. According to the developed methodology, the hydrological and hydraulic assessment with the structural evaluation formed the Level 1 assessment. Depending on the results of Level 1, the bridge owner is directed to the Level 2 assessment, if necessary, for a detailed bridge scour analysis and numerical modelling of flow and sediment transport. In this study, only Level 1 was conducted.

Hydrologic Assessment of Watershed Characteristics
Hydrological assessment of a watershed with the outlet located at river bridge section was performed using several hydrological indicators, such as morphometric index, watershed land use and vegetation cover, physical structure, watershed precipitation magnitude and type, landslide and flood history, and water detention/retention structures in the watershed, as given in Table 2. Morphometric index (MI), a parameter, was introduced to estimate the flash flood potential of the watershed using selected morphometric parameters, such as bifurcation ratio (Rb), drainage density (Dd), stream frequency (Fs), texture ratio (T), ρ coefficient, length of overland flow (Lo), elongation ratio (Er), shape factor (Sf), form factor (Ff), compactness coefficient (Cc), circularity ratio (Cr) and width to length ratio (WLr). MI was calculated as the average of the ranks of those morphometric parameters. Other hydrological indicators were complementary features affecting rainfall-runoff process and soil erosion. Then, a hydrological grade (HG) was introduced for rating the overall watershed hydrology and flash flood potential of that watershed. HG was computed by summing

Safety-Inspection Study
The current safety-inspection method for bridges in Turkey lacks the consideration of the functionality of the bridges. Therefore, the developed methodology was especially designed for river bridges in which various hydrologic and hydraulic factors were considered. As Johnson [27] pointed out that overall evaluation of river bridges should include stability indicators, starting at watershed characteristics or vertical and lateral aspects of the channel; the current method involved four main parts, all concentrated on hydrological and hydraulic factors. The first part involved a hydrological assessment of the watershed and flood potential while the second and third parts consisted of the stability indicators of stream and bridge characteristics, respectively, that were determined or modified using past studies in the literature, lessons learned from failures of river bridges in Turkey, and observations from field studies. During evaluation of the indicators, characteristics of the indicator in the upstream and downstream parts of the bridge were assessed. In the last part, a rapid scour check was carried out to determine scour susceptibility of the bridges. Besides hydrological and hydraulic assessment, a structural evaluation of the river bridges was also included, which was composed of evaluation of main body, and earth retaining and serviceability components of bridges in regard to loss, damage, aging and deformation. According to the developed methodology, the hydrological and hydraulic assessment with the structural evaluation formed the Level 1 assessment. Depending on the results of Level 1, the bridge owner is directed to the Level 2 assessment, if necessary, for a detailed bridge scour analysis and numerical modelling of flow and sediment transport. In this study, only Level 1 was conducted.

Hydrologic Assessment of Watershed Characteristics
Hydrological assessment of a watershed with the outlet located at river bridge section was performed using several hydrological indicators, such as morphometric index, watershed land use and vegetation cover, physical structure, watershed precipitation magnitude and type, landslide and flood history, and water detention/retention structures in the watershed, as given in Table 2. Morphometric index (MI), a parameter, was introduced to estimate the flash flood potential of the watershed using selected morphometric parameters, such as bifurcation ratio (R b ), drainage density (D d ), stream frequency (F s ), texture ratio (T), ρ coefficient, length of overland flow (Lo), elongation ratio (E r ), shape factor (S f ), form factor (F f ), compactness coefficient (C c ), circularity ratio (C r ) and width to length ratio (WL r ). MI was calculated as the average of the ranks of those morphometric parameters. Other hydrological indicators were complementary features affecting rainfall-runoff process and soil erosion. Then, a hydrological grade (HG) was introduced for rating the overall watershed hydrology and flash flood potential of that watershed. HG was computed by summing up the ranking values (RV) of the five indicators given in Table 2. Since the RV of each of these five indicators ranged between 1 and 12, HG can have a minimum of 5 and a maximum of 60.

Indicators of Stream Stability
Channel cross-section was subdivided into three groups as bed, floodplain and banks, to determine vertical or horizontal stability of the stream. River training structures, material type and cover, formation, and sediment dynamics in the channel sections were considered in this section. Most of the indicators and their classifications were taken from Johnson [27] with small adjustments to their limits depending on the field observations in Turkey for the calibration of the classification of the indicators.

Stream Bed Characteristics
Stream bed characteristics consisted of flow habit, stream longitudinal profile, channel pattern, bar development, bed material and protection, as given in Table 3, in which S is the slope of the stream, F s is the approximate portion of sand in stream bed, and w and y are the stream width and flow depth, respectively.
Although flow habit may be considered as a hydrologic response for stream flow, discharge is mainly an interaction with stream bed. Therefore, this indicator was taken into consideration in this part. For instance, Karaman Bridge in the project area was located on an ephemeral stream and failed due to heavy rain in the dry season. It was also observed that flow habit was seriously affected by hydroelectric power plants (HEPP) and dams in the area, since they could hold or suddenly release water during their operation. Moreover, indicators of stream longitudinal profile, channel pattern, bed material and bar development directly denote flow depth and velocity, sediment transportation mechanism and capacity. Another indicator was chosen as the bed protection structures which might contribute to the grade control of stream bed. However, they may cause accumulation or degradation at low flow periods. For instance, grade control structures located nearby Filyos I Bridge in the pilot area caused huge scour holes and instability of bed in vicinity of the bridge.

Floodplain Characteristics
The formation and width of floodplains were selected as indicators as given in Table 4. These indicators were used to assess the capability of floodplains for carrying excess bankfull discharge. Occupation of these floodplains by some residential units is a common problem in Turkey. This leads to increase in bed-roughness conditions. Therefore, even during moderately frequent flood events, the river flow rises such that the nearby lands become inundated. This problem was also observed in the pilot area, especially near settlements. For instance, the floodplain in the downstream part of Bartin III Bridge was occupied by local people for private use near the stream. During February 2016 flood, this area was entirely inundated. The degree of damages depends on the intensity of all types of properties located at the floodplains.

Stream Bank Characteristics
Stream banks are the most important components of a channel cross-section for the assessment of the vertical stability of streams. Bank height, average bank slope, vegetative bank protection, soil texture, mass wasting, or bank failure and bank cutting are the dominant indicators for the stability assessment of stream banks, as given in Table 5.
Instability in stream banks may cause lateral bank erosion, and deepening or widening of channel and floodplain developments due to slumped materials in the bank. Evidence for potential instability for channel banks should be investigated during inspection of river bridges. In this part of the method, the bank stability assessment was adopted from Johnson [27], which was thoroughly studied and tested, while field observations also approved its competence.

Other Effects
During field studies, it was observed that some human-induced activities and site-specific channel characteristics resulted in instability problems in vicinity of river bridges. Hence, these types of activities were considered under this heading. These were gravel mining from the stream bed, the presence of stream conjunctions, debris jamming/accumulation, river reclamation, percentage of channel constriction and new bridge construction near the current bridge, as given in Table 6.
Gravel mining from the stream bed is a common problem, especially in the Western Black Sea Basin. There exists a circular issued by Turkish State Highways, which states that gravel mining from stream beds cannot be performed at 750 m in upstream and 1000 m in downstream of the river bridges. However, many bridges in the area are still affected by this practice. For instance, it was observed that a significant amount of bed material is extracted uncontrolled at a site 2 km downstream of Gokcebey Bridge.  Another important factor was chosen-the stream conjunction, which affects bed slope of the main channel by carrying and accumulating significant amounts of sediment at those locations. Therefore, in the study, ruggedness number (R) described by Melton [43] was considered to describe the hydro-geomorphic process of floods to debris floods [44]. Another important indicator is debris accumulation at piers and in the openings of river bridges, which can easily be spotted during low flow seasons. During field trips, it was observed that accumulated debris, especially around piers, was not cleared causing an increase in scour hole depths. Another important issue was river reclamation, which is a common practice in the area due to frequent flood events in the basin. However, observations showed that it might change flow structure and cause channel migration in the vicinity of bridges. Hence, extra care should be given when a river bridge is involved in the process. During project duration, river reclamation was conducted near Gokcebey Bridge and it was seen that alignment of water was diverted to a pier. This is an important issue, since am increase in lateral channel migration after reclamation may accelerate the failure of river bridges, as stated by Johnson [27]. It was also observed that new bridges constructed near the old ones might cause increases in the local turbulence fields and scouring. Therefore, this was taken as an indicator in the developed method.

Bridge Characteristics
Some characteristics of river bridges may increase local scouring around foundations. As given in Table 7, bridge location, upstream distance to bridge from meander impact, bridge skewness, flow in the spans, and pier and abutment characteristics were selected as indicators of bridge characteristics. During inventory review and field studies, it was noted that location of bridges was an important common feature, besides distance of bridge site from a meander impact point and bridge skewness. A detailed review of the inventory of General Directorate of Highways showed that the skew angle of river bridges can increase up to 67 • due to topographic or economic constraints. Most of the known failed bridges were observed to have a skew angle of more than 10 • . Among inspected bridges, Filyos V Bridge had a skew angle of 45 • while the skew angle of Bartin I Bridge was 20 • . In an experimental study, Demir [45] showed that a bridge deck model with a skewness of 15 • increased the maximum depth of scour up to 57% under submerged flow conditions. Another important feature is the spans of the bridges that should be larger, and not constricted due to sediment deposition or debris accumulation, to let excessive water to pass through, especially during floods. Otherwise, submerged flow conditions, increased flow intensity and higher turbulence may result in excessive scouring around piers and the bridge may need to be closed to traffic for public safety resulting in additional operational and time costs. For instance, Bartin I Bridge was inundated during a 1998 flood event and the abutment of Gebelit Bridge failed in a 2013 flood event due to low meandering impact distance, bridge skewness and submerged flow conditions. Channel alignment, geometric characteristics and observability of piers, degree of scouring and existence of countermeasures around piers are also important indicators of stream stability and local scouring. Observability of bridge piers is also considered, since under water inspection it would be necessary to check the degree of scouring around piers. Moreover, inspection studies are easier around abutments in low flow seasons. However, in some cases, environmental factors affecting abutments and approach fill conditions should also be examined in terms of vulnerability of abutments with respect to erosion.

Rapid Procedure for Scour Check
In addition to the hydrological and hydraulic evaluations, it is recommended to perform a rapid and simple scour check, specifically to detect any potential scouring problem at the foundations. Thus, river bridges that are not ranked as poor based on the developed method, but are still scour critical can be identified. The scour check involves information about foundations, the presence of any countermeasures against scouring, and characteristics of the river flow and bridge characteristics, such as skew angle, which directly increase scouring around infrastructural elements. Depending on the result of this assessment, a more detailed scour risk assessment could be advised in Level 2, including empirical equations [46,47] and numerical model simulations to predict hydraulic conditions and sediment transport around piers and abutments. This level of assessment is recommended if the bridge is considered to be poor in Level 1, the basic assessment of the bridge.

Structural Assessment of River Bridges
Elements of river bridges should be inspected with regard to structural stability as well, every two years [48]. Berk [49] and Yanmaz et al. [50] noted the necessity of periodic inspections and effective factors on inspections and formed a structural checklist for river bridges in Turkey using sample bridges in USA and Turkey. That structural assessment was composed of main body, earth retaining and serviceability components of bridges. In this study, a ranking of the checklist for the structural evaluation of river bridges in Turkey proposed by Yanmaz et al. [50] was modified using the same classes as in hydraulic assessment. Tables 8 and 9 show the indicators of structural assessment and their classifications, respectively. Bridge characteristics specified in Tables 8 and 9 were obtained via visual inspection or from design projects of the bridges. Figure 2 illustrates the main steps of the developed safety-inspection algorithm. The method used for classification of the indicators was adopted from Johnson [27]. Each indicator of the watershed hydrology and flash flood potential, stream stability and bridge characteristics in the safety inspection algorithm was classified as excellent, good, fair and poor, and a rating value was assigned for that indicator ranging between 1 and 12. Johnson [27] noted that weighted indicators yielded the same results as the equally weighted indicators.

Evaluation of Scoring of the Method
Therefore, in this study, each indicator in the method was also equally weighted as the rating value was determined. Then, a risk point (RP) was introduced in order to make a decision using the grading of the indicators. According to the adopted method, the initial value of the RP was set to zero and then for indicators classified as excellent or good, the value of RP did not change; while for indicators classified as fair, RP was increased by one; and for indicators classified as poor, RP was increased by two. Table 10 shows the change in RP with the grades of the indicators of hydrological and hydraulic assessment. The minimum and maximum values of HG changed between five and 60, respectively, as seen in Table 2; and the ranges of HG were determined by dividing the highest possible difference in the grades of HG into four equal parts for the classification groups of excellent, good, fair and poor. Then, starting from five and taking excellent and good classes as one, since RP was not changed for those classes, the ranges of HG were determined. Thus, when all the indicators were classified as excellent or good, the RP would be zero, and if all the indicators were classified as poor, the RP would be 66, since there was a total of 33 indicators in the method (one for watershed hydrology, six for a stream bed, two for a floodplain, six for a stream bank, six for other effects and twelve for bridge characteristics). Then, the bridges were ranked according to the ranges of RP given in Table 11. The ranges of RP were determined by dividing the highest possible value of RP into three equal parts, each corresponding to the computation steps of RP: no change, increased by one and increased by two. The first part was then divided into two to determine the two-rating classification as excellent and good. This approach was adopted in the method, since the RP was not changed for indicators classified as excellent or good. A structural grade (SG) value, RP, was introduced and computed from the sum of RV of all indicators of structural assessment. According to the method, the SG was classified as given in Table 12 and the classification of its ranges was determined similarly to the RP. At the end of the assessment, a rapid scour assessment was carried out.  Listed in Table 8 New or undamaged, damaged or deterioration is yet in its early stages Slightly damaged but noticeable, causing no problem Damaged but still functioning, stability problem has started to occur Severely damaged, serious safety problems, urgent action needed, element deficiency or cannot be assessed Therefore, in this study, each indicator in the method was also equally weighted as the rating value was determined. Then, a risk point (RP) was introduced in order to make a decision using the grading of the indicators. According to the adopted method, the initial value of the RP was set to zero and then for indicators classified as excellent or good, the value of RP did not change; while for indicators classified as fair, RP was increased by one; and for indicators classified as poor, RP was increased by two. Table 10 shows the change in RP with the grades of the indicators of hydrological and hydraulic assessment. The minimum and maximum values of HG changed between five and 60, respectively, as seen in Table 2; and the ranges of HG were determined by dividing the highest possible difference in the grades of HG into four equal parts for the classification groups of excellent, good, fair and poor. Then, starting from five and taking excellent and good classes as one, since RP was not changed for those classes, the ranges of HG were determined. Thus, when all the indicators were classified as excellent or good, the RP would be zero, and if all the indicators were classified as poor, the RP would be 66, since there was a total of 33 indicators in the method (one for watershed hydrology, six for a stream bed, two for a floodplain, six for a stream bank, six for other effects and twelve for bridge characteristics). Then, the bridges were ranked according to the ranges of RP given in Table 11. The ranges of RP were determined by dividing the highest possible value of RP into three equal parts, each corresponding to the computation steps of RP: no change, increased by one and increased by two. The first part was then divided into two to determine the two-rating classification as excellent and good. This approach was adopted in the method, since the RP was not changed for indicators classified as excellent or good. A structural grade (SG) value, RP, was introduced and computed from the sum of RV of all indicators of structural assessment. According to the method, the SG was classified as given in Table 12 and the classification of its ranges was determined similarly to the RP. At the end of the assessment, a rapid scour assessment was carried out.

Results and Discussion
Five river bridges located in the Western Black Sea Basin were selected for detailed inspection and the application of the proposed method. Results of ratings of those bridges with regard to hydrological and hydraulic bridge characteristics are given in Table 13. Starting from zero, the RP value was determined, as given in Table 13, for both hydrological and hydraulic evaluation. In the table, the overall score of hydrological evaluation HG was obtained using Table 2 and was classified as given in Table 9. Furthermore, for each bridge the single underlined RVs in the column increased RP by one while double underlined ones increased RP by two. For structural assessment, the RV of each indicator in the same column presented in Table 14 was simply summed up to obtain SG values for each bridge. Table 15 shows the HG, RP and SG values of those bridges inspected and Table 16 gives the grading of the bridges with regard to hydrological, hydraulic and structural indicators. Finally, a rapid scour assessment was performed as given in Table 17.
As seen from Table 15, the most critical river bridge was determined to be the Filyos V Bridge, while Bartin III was found to be the least critical, based on hydraulic assessment. In order to interpret the results and evaluate the capability of the method, available inventory of those inspected bridges was examined. It was found that Filyos V Bridge was repaired after a 1998 flood event due to a settlement of 65 cm at one of its piers, and since then the bridge was kept under regular inspection due to the persistence of the problem. After some maintenance work for the solution, the 15th Regional Directorate of GDH, the local bridge owner, decided to demolish the bridge. However, the bridge is currently in service due to the lack of alternative routes. Therefore, possible alternatives are being considered for the solution. Contrarily, Bartin III Bridge, which was determined as the least critical with respect to hydraulic assessment, has not ever been submerged or closed to traffic in its service life even during the catastrophic flood of 1998, which affected Bartin I Bridge and many others in the area. The rest of the inspected bridges were ranked "fair," as given in Table 16.
Another example for effective hydraulic factors was Gokcebey Bridge, which had a crack at its abutment and experienced serious maintenance in 1968. Then, again in 1999, it was reconditioned after a progressive settlement at one of its abutments. Repeating similar problems at those bridges show the importance of detecting problems caused by hydraulic factors, since the condition is persistent and only defining the source of the problem can enable an appropriate solution. When the results of the application of the method were analyzed, it was found that the bridges having problems due to hydraulic factors were successfully identified. Moreover, further assessment for Filyos V Bridge was strongly recommended according to the rapid scour assessment, while scouring around piers of other bridges was considered negligible, as presented in Table 17. It was also seen that there was a good agreement between the results of the proposed method and the rapid scour assessment, showing that indicators of hydraulic factors affecting scouring around piers were successfully identified and ranked. According to the structural assessment, the most critical bridge was found to be Bartin III Bridge (Tables 15 and 16). During visual inspections, serious structural problems, such as segregation, deterioration of concrete cover leading to corrosion of reinforcements, and cavities on the concrete surface at the bridge were observed. Bartin III Bridge got the highest SG value from the inspection. Therefore, the safety-inspection method classified the bridge as structurally poor. Hence, it was recommended that Bartin III Bridge should urgently be inspected with equipment and necessary maintenance should be carried out.
Considering the current situation of Filyos V Bridge, some amendments could be suggested with simple or comprehensive measures that can be made on the auxiliary elements of the bridge, which was classified as the poorest bridge among four other bridges. Generally speaking, scour holes observed at the piers were assumed to be partly induced by debris accumulation, such as tree branches and trunks. A simple scaled rod can be installed at scour critical piers of the bridges, thus scour depth around piers can be observed. It should be noted that important bridges crossing large rivers may be supplemented by a sonar instrument for continuous monitoring of scour patterns. Sufficient quantities of riprap can also be dumped into scour holes to mitigate further scouring around piers of the bridge. Damages to the guiding wall of abutments and approach fill should be repaired. That way, a reduction in the RP of the bridge by 11 points can be achieved. Thus, instead of a costly solution, such as reconstruction of a new bridge, the risk factors can be reduced by repairing the problematic parts. Moreover, construction of a new bridge without considering the stream stability and other local hydraulic indicators may result in the same problems in time. Therefore, inventory of those bridges and lessons learned can be used during design stage of new bridges nearby. Furthermore, HYRISK [51], a risk-based software can be applied to determine the annual risk of scour to a bridge, as done by Yanmaz and Apaydin [52] for several bridges in Turkey. Bartin III Good Gokcebey Good Age of the bridge is an important indicator for hydraulic/hydrologic and structural aspects. The effect of the age is implicitly considered in this study for structural evaluation of bridge components. Based on in-situ observations and some tests, e.g., aiming to detect the instantaneous compressive strength of concrete, the aging effect can be considered. Using such information, the approximate remaining lifetime of the bridge concerned can be estimated. Caner et al. [48] developed a simple method to assess the remaining service life of a bridge through defining a relationship between its current condition rating and its age by evaluating a set of bridges of different ages in Turkey. With the integration of this method, more realistic grading can be handled for structural components of bridges. The age of the bridge can also be considered for a more comprehensive analysis of the scouring history. However, with no field data reflecting scour hole geometries belonging to individual past storms, a realistic analysis of the effects of past storms on bridge scouring cannot be handled. Even limited scour data are available for some bridges; accuracy of the data is highly questionable, since it is almost impossible to gather scour data during the period of maximum flows. When the scour depths are measured after the flood wave leaves the bridge section, the scour hole might have been filled with the sediment accumulated during the falling stage of the flood or debris accumulation. Therefore, maximum scour depth data are not available. Furthermore, even if highly sophisticated sonar sensors are installed on bridge piers for continuous monitoring of the streambed, they may be swept away by the high impact effect of flood wave. Therefore, the valuable information will be lost. On the other hand, it is a regular approach by the GDH to fill the scour holes around piers and abutments of bridges with riprap after heavy floods. Carrying that out would certainly affect local turbulent flow field, hence the bed regime. As a concluding remark, although past flow data may be available, the corresponding conditions at bridge sites are definitely unknown. This makes it impossible to attain a reliable scouring history for a bridge. Moreover, it should be kept in mind that the proposed inspection check is based on instantaneous field observations reflecting the date of visit at several seasons to consider the effect of variations in flow conditions. With the detailed studies concerning evaluations of geomorphic variables and hydrological information, more realistic information can be obtained for flood characteristics, which would definitely characterize the type and depth of scour hole, to be used in Level 2 analysis.
When the proposed algorithm is accepted to be used as an integral component of the BMS by GDH, bridges will be assigned a kind of identity card. With the implementation and use of the proposed algorithm, which is partly based on in-situ observations and hydrological evaluations, it will eventually be possible to rank the bridges according to priorities in regard to repair and maintenance needs. With the collected future information and its interpretation, more robust validation would be achieved for those bridges, which were identified as critical ones.
Globally, there are numerous BMSs, ranging from very sophisticated ones with detailed measurements and advanced decision-making systems, to very simple ones with only basic visual inspections. Features of the BMS depend on the characteristics of the bridge stock, the need and future projections of the country and its budget. For instance, Pontis is used in the USA [53] with a well-developed and comprehensive database; namely, the National Bridge Inventory. The UK, which has one of the oldest infrastructural systems in the world does not have a national BMS. It has a Structures Information Management System, two manuals for design, assessment and operation, with one manual for guidance on scouring on infrastructure. On the other hand, some countries, such as Belgium, France, Germany and Ireland use engineering judgement for decision-making instead of a BMS [7]. In the BMS introduced in Turkey in 2017, the hydrological and hydraulic factors were not sufficiently considered for river bridges. Therefore, the developed-safety inspection method that could be implemented into the BMS would enable a comprehensive database for river bridges considering the work load and budget limits of the GDH.

Conclusions
In this study, a safety-inspection method was developed for river bridges in Turkey. The method specifically takes into account the effects of hydraulic factors on river bridges. The Western Black Sea Basin, Turkey, was selected as the study area, where incidents of severe damage or failure of river bridges frequently occur. Inspection methodology consists of watershed hydrology and flash flood potential, and stream and bridge characteristics, with a rapid and simple scour assessment. Most of the indicators about stream stability were taken from Johnson [17], lessons learned from past failures of river bridges and observations from field studies. Due to the absence of required data for calibration of the classification of the indicators and the method, five river bridges in the study area were chosen for detailed inspection and observation. Results of hydraulic assessment showed that Filyos V Bridge was ranked as fair, with the highest rating score among the sample, while Bartin III Bridge was good and the rest were classified as fair. Inventory of the bridge also proved that serious problems due to hydraulic factors were encountered and demolishing of the bridge was an option. A rapid scour assessment also recommended that Filyos V Bridge was a scour-critical river bridge and further studies should be undertaken. Although Filyos V and Gokcebey Bridges were on the same watercourse and close to each other, Filyos V Bridge was found to be more critical. This might show the importance of bridge characteristics that were taken into consideration. It was also noted that gravel mining was a serious problem in the area. This was a threat, especially for Gokcebey Bridge, downstream of which is a nearby runoff river plant. During the three-year inspection period, debris at the piers was not removed and found to create a serious problem for increasing scouring potential around the piers.
Bartin III Bridge was found to be subject to structural deficiencies of main body components. Therefore, retrofitting of the bridge should be performed.
It should be noted herein that the calibration of the classification limits of the indicators and the method could be performed with the collection of necessary input data from numerous river bridges over time, thus the capability of the developed algorithm could be improved. With the help of the developed safety-inspection method, the management and maintenance of river bridges in Turkey could be improved, and necessary maintenance work could be scheduled according to the RP values of the bridges. Thus, the lifespan and the serviceability of the bridges could be extended. Furthermore, new bridges that could be exposed to increased flood potential can be designed accordingly. A failure risk assessment using those indicators as inputs of channel, substructure and scour criticality conditions, can also be estimated. Developed by considering hydrological, hydraulic and management conditions in Turkey, this method can be recommended for use in countries with no detailed safety-inspection methodology.