Bridge Element Deterioration Rates

TRANSPORTATION RESEARCH RECORD 149 9 Bridge Element Deterioration Rates lmad J. ABED-AL-RAHIM AND DAVID W. JOHNSTON Prediting the deterioration rates of bridge elements is an important?ponent of any ridge management system. This is beause the pred1t1n of future bndge funding needs is based in part on the existing ad. future ondtion of the bridge element. A methodology for pred1tmg the detenorat1on rates of bridge elements was developed on the basis of an analysis of historial data from bridge inspetions. The methodology is applied to the bridge dek, superstruture, and substruture as example elements. General deterioration urves were developed for the three major bridge elements by material type. More detailed deterioration urves for the bridge elements were also developed for various subgroupings of these elements divided by material and environmental fators. Deterioration is the proess of deline in bridge element ondition. It is aused by the environment, traffi, and other spontaneous fators. The predition of future bridge funding needs is made in part on the basis of the existing and future onditions of the bridge element. It is thus important for the suess of any bridge management system to aurately predit the bridge element deterioration rates. Under urrent FHW A inspetion proedures elements (suh as the dek, superstruture, and substruture) are evaluated on a sale of 9 to indiating the degree of deterioration. Unless maintenane or rehabilitation work is performed on the bridge, the element on-. dition rating would be expeted either to remain unhanged or to drop in any inspetion period. The inspetion of bridges is onduted by trained tehniians under engineering supervision every years. Bridge-owning agenies keep reords of the onditions of the various bridge elements in the Bridge Inventory data file along with other bridge data. Aording to the FHWA's Bridge Management Systems report (1) all studies to date on bridge deterioration rates tend to predit slower delines in bridge ondition ratings after 15 years or so. The report also inluded results from a regression analysis of National Bridge Inventory (NBI) data for deterioration of dek ondition and overall strutural ondition. The results suggest that the national average dek ondition rating delines at the rate of.14 points per year for approximately the first 1 years and.5 points per year for the remaining years. For overall strutural ondition the values were.94 per year for 1 years and.5 per year thereafter. This implies that the average onditions never fall below a ondition rating of 6 until after 6 years. However, these results do not fit with the experiene enountered in pratie, whih suggests a muh faster deline in ondition. The primary diffiulty enountered by researhers in developing a reasonable representation of the atual deterioration urves is that the models used to analyze aggregate inventory onditions at a point iri time did not take into aount the effets of any improvement work done to the bridge elements in the past. Department of Civil Engineering, North Carolina State University, Raleigh, N.C. 695. OBJECTIVE The objetives of the study were to develop analytial methods for estimating the deterioration rates of the three major bridge elements (dek, superstruture, and substruture) as a funtion of material types and various environmental fators. The mathematial method developed was to allow periodi reanalysis by using existing (but then urrent) North Carolina Department of Transportation (NCDOT) bridge data bases. LITERATURE REVIEW Several efforts have been made. to estimate the deterioration rates of bridge elements. A study onduted at the Transportation Systems Center (TSC) () used NBI data and regression tehniques to develop equations that related the three major bridge element ondition ratings to other bridge harateristis found in the NBI. The study inluded only bridges that were 5 years or younger. Age was found to be the most highly orrelated fator, with average daily traffi (ADT) being the next highest. The equations developed were used to predit the hange in bridge ondition over time. It was suggested on the basis of the equations that were developed that the dek deteriorates slightly faster with age than the superstruture or substruture. The study estimated the average deterioration of deks to be about l point in 8 years and that of both the superstruture and substruture to be about 1 point in IO years. The FHWA's Bridge Management Systems report (J) indiated three weaknesses in the TSC study. The first was that the analysis was performed on bridges that were no more than 5 years of age. The seond was that the equations developed assumed linear relationships between the bridge element ondition ratings and the parameters inluded in the equations. The third weakness was that the interept oeffiient in the equations was onstrained to 9. In a study by Hyman et al. (3) for the Wisonsin Department of Transportation pieewise linear regression was used on numerial ondition appraisal data to develop deterioration urves. The study estimated a omposite deterioration urve for all bridge types. In addition, deterioration urves were developed for six different bridge types: steel dek girders, other steel strutures, reinfored onrete dek girders, onrete slabs, prestressed onrete strutures, and ulverts. Chen and Johnston ( 4) onduted a survey of bridge inspetors and maintenane supervisors to determine age to the various levels of ondition on the basis of aumulated expert experiene by a Delphi approah. A series of trilinear deterioration relationships was developed, largely on the basis of survey results, for major bridge elements and material types. Jiang and Sinha (5) used two approahes for developing deterioration urves. These were (a) regression analysis of ondition versus age and (b) Markov hain model tehniques.

IO TRANSPORTATION RESEARCH RECORD 149 The Markov hain model tehnique was used for two kinds of preditions: the ondition rating of a bridge at a given age and the servie life of a bridge. This tehnique was based on defining states in terms of a bridge element ondition transiting from one ondition to another. The zoning tehnique was used to obtain transition matries, sine the rate of deterioration of bridge onditions varies at different ages, thus making it a nonhomogeneous proess. "Bridge age was divided into groups and within eah group the Markov hain was assumed to be homogeneous" (5). A transition probability matrix was therefore developed for eah group. The Markov hain approah produed unusually slow preditions of element deterioration in omparison with those produed by Chen and J ohnsto n' s ( 4) surveys and in omparison with the subjetive experiene. However, the urve shapes, a flat S urve, were similar to the trilinear shapes developed by Chen and Johnston (4). Saito and Sinha (6) also used the Delphi approah to develop deterioration urves for the diffeent bridge elements. This was based on a survey of 14 Indiana Department of Highway employees in harge of bridge inspetion and design. FHWA's Bridge Management Systems report (J) stated that "All the studies on bridge deterioration to date imply that the rate of deterioration tends to slow down markedly after 15 years or so. In fat, data from many studies-when taken at fae value-suggest that the average bridge ondition atually improves or heals with age at some point. This is due to the fat that in most of the studies mentioned earlier no onsideration was given to the effet of the work performed on the bridge ondition rating. Suh effets.will mask the atual relationship between the bridge's age and the element's ondition rating. DATA ON BRIDGE ELEMENT CONDITION RATINGS FHW A requires bridge-owning agenies to keep reords of numerous harateristis for every bridge under their jurisdition. Element ondition ratings of the dek, superstruture, and substruture are part of these reords. NCDOT has been keeping suh reords sine 198. These data, whih are updated as new inspetions our, are kept in the North Carolina Bridge Inventory (NCBI) data file. Astatus reord of the total file is retained at the end of eah fisal year. Seleted fields of these reords, inluding bridge element ondition ratings, are stored in the Bridge History files. These files are appended annually to inlude reords of the latest fisal year. Unless it is reorded as an N, for nonappliable, the bridge element ondition rating an only be an integer from Oto 9. Thus, when a bridge element hanges from one ondition rating to another it an only hange in integer values suh as 1 and. Hene, the data for ondition rating versus time do not yield a urve when they are plotted (Figure 1). Bridge elements almost never reeive ondition ratings, 1, or beause they are either rehabilitated or replaed before they reah suh onditions. Of more than 14, bridges in North Carolina, eah with three primary elements, only one bridge element had a ondition rating of and none had a ondition rating of or 1 in 1989. A bridge element only rarely reeives a ondition rating of 3 sine, one again, they are generally either rehabilitated or replaed before reahing this level. Only 185 bridge elements in North Carolina were rated at a ondition of 3 in 1989, and the majority of these were timber bridge elements. 9-8 6 - Cl : 5 : g '5 : 4 3 FIGURE 1 Time Condition rating versus time. PROBLEMS RELATED TO BRIDGE CONDITION RATINGS FHW A provided a oding guide () for evaluating the ondition ratirigs of the various bridge elements. However, it did not provide a detailed referene guide that would expliitly desribe the relationship between the deterioration levels of the bridge elements and numeri ondition ratings (6). Thus, what might be reorded by one inspetor as a 6 might be reorded by another inspetor as a 5. An atual measure of the effet of this phenomenon on the onsisteny of the data stored is hard to measure. The states and FHW A have, however, attempted to promote onsisteny through inspetion training. When work improvement is performed on the bridge, it will inrease the ondition of the bridge but it will not affet the age, thus distorting the atual relationship between age and ondition rating. Although NCDOT keeps reords of all of the work performed on the bridges, it is diffiult to measure the ontributions of various improvement ativities toward the ondition of the bridge elements. This is aused by the fat that members of a rew performing one type of repair work might go ahead and perform some other minor repairs to other omponents of the bridge while they are at the site. The ondition rating of the other omponents might thus improve, although the work might inorretly be reorded only under the primary work item ode. Work improvements performed on bridges will in general either improve the ondition rating of the element or inrease the stay of the element in its urrent ondition rating. Suh work will thus disrupt the atual relationship between the bridge age and ondition rating. This is evident in Figure, in whih the average age of bridges with ondition ratings of 4, 5, and 6 are almost equal, whereas the average age of bridges with a ondition rating of 3 is less than the average age of the bridges with the previous three ratings. It an also be seen from Figure that there is typially a lot of variation in the data for bridges with the different ondition ratings. These problems are aused by looking at data from only I year for bridges with no previous work, bridges with some previous work, and bridges that may have been substantially rehabilitated in the past. Unfortunately, sine reords only extend bak to 198 for some data and even less for other data, these groups of bridges an-

Abed-Al-Rahim and Johnston II A = 1 obs., B =,..., Z => 6 9 A 8 WZZZHDMGEHEGMFFFAAIMKRPZNLLKLJJIOA B Average age of bridges at ondition rating AA A A A AA C> EIJRPIMI<MPSOSSUZLVZZZZZZZZZZZZXZZZMFA A C A A B B A : a; : :g (.) CJ) 6 CGPOEHEHGGJMTMVZZZZZZZZZZZZZZZZZZZHFC D C BAABBAA A AAA BA A A AA A B A A 5 EGGKBACBD DCBEJJKRZZZZZZZZZZZZZZZZJBH A BB AA AB AB B. B 4 ED A A ACBA A DCBGOEGORPZNNIQPNKAB AA AA A A A A A A A A 3 A A A A ABBBCECBCADBAABB A A A 5 1 15 5 3 35 4 45 5 55 6 65 5 8 Age (Years) FIGURE North Carolina timber dek ondition ratings versus age. not be separated. A method other than regression on data from a single year must be found. CHARACTERISTICS OF BRIDGE CONDITION RATING DATA The atual relationship between a material ondition and time should yield a ontinuous urve when plotted. However, the shape of the "urve" will be affeted among many things by the definition of the ondition ratings. Take for example a ase with two different sales. The definition of a 9 rating might be the same in both sales, but the definition of an 8 rating is very good ondition in one sale and average in the other. The time that it takes for a bridge element to drop from a 9 to an 8 will therefore be different for the two sales. Furthermore, the "urve" representing the relationship between the ondition rating and time for the two sales will be different. The urves will also be different in the ase in whih two sales have different ranges. An example of this would be the sale used by FHW A, whih has ratings from to 9, versus the one used by Saudi Arabia, whih has a range of to (8). As mentioned earlier, aording to the FHW A definition of ondition ratings, the data will not yield a "urve" when plotted against time (Figure 1 ). However, a urve an be plotted if the slope of various ondition ratings inrements an be estimated. A more realisti urve would be one that has a series of linear segments between the suessive ondition ratings (Figure 3). The only time measurement related to the ondition rating that an be diretly used in suh analysis is the age of the bridge. However, for traditional tehniques suh as regression and the Markov hain model, this relationship is usually distorted by previous work improvements performed on the bridge elements. It was therefore neessary to develop a methodology by whih the relationship between the bridge element ondition rating versus time ould be analyzed. PROPOSED DETERIORATION ANALYSIS. METHOD The approah proposed for finding a solution is to onsider eah ondition rating separately. One a ondition rating (r) is hosen, 9 8 O> : 6 : 5 +:: :a 4 : (.) 3 FIGURE 3 Linear segments onneting suessive ondition ratings.

1 TRANSPORTATION RESEARCH RECORD 149 11 1 9 en 8 : : Cl> en.so.. 5 Q). 4 E ::::> z 3 n r,o N "'" +n +... +n r r,o r,1 r,m 1 n r,m-1 n r,m m-1 m Number of Points Delined in One Year FIGURE 4 Number of observations of ondition.rating deline. bridges with that ondition rating are identified from the Bridge History file for.a seleted year, plant. Reords of the element ondition rating of the identified bridges for the following year, t + 1, are then ompared with r. Initially, bridges were onsidered if the ondition rating in the following year, t + 1, either did not hange or delined to a lower rating. This was to eliminate improved bridges from the study. - The total number of bridges, N" having a ondition rating of r in year t an be tabulated. For example, Figure 4 shows a distribution of the number of bridges from a partiular subset that either hanged by points (i.e., did not deline) or delined to a lower ondition rating by I year later, t + I. The number of bridges for whih the ondition rating hanged by j points from the original r is represented by n,. 1 with m being the maximum deline possible for r. The summation of n,..j for all possible j's will thus be egual to N,. The average weighted hange within that I-year period seleted will be equal to Ill I n,.j Xj A VGCHN, = 1 - =_o (1) N, where AVGCHN, =average hange from ondition rating r within the I-year period seleted (t, t + I); n,_ 1 = number of bridges hanging by j points from ondition rating r; j = r 1 - (element ondition rating of the same bridge in the following year); m = maximum number of points the bridge element an drop from r; and N, = total number of bridges at. r.in year t, The time that it takes to drop by 1 point from r to r - I an thus be alulated one A VGCHN, is determined by using similar triangles (Figure 5): TIME, r - (r- 1) () (t + I) - t AVGCHN,. where TIME, is the time that it takes to drop by 1 point from ondition level r. Equation_ an be redued to 1 (3) TIME, = A VGCHN, en : g 'i5 : (.) -VCHI -'----..-.r1 r-1 - - - - - 1 - - - - r - i.-rtme I I r I I I I I YRt YRt+ 1 Time FIGURE 5 Condition rating versus time for seleted r.

Abed-Al-Rahim and Johnston 13 New Element 9-8 O'> "+:: ell 6 5 4 :t:: 13 () 3 Rehabilitation 1 FIGURE 6 Time Effet of work improvement on bridge element ondition rating. Equation I an be modified to the following form for use when data exist for multiple 1-year intervals: AVGCHN, = where YRL-1 m L L ntr.r.j> Xj t=yri j=i YRL-1 m L L n(r.1,j) t=yri_ j=o YR I = first year seleted, YRL = last year seleted, and t = year being onsidered. The equation an be applied for eah value of ondition rating r, alulating the slopes for the linear segments onneting suessive ondition ratings. Plotting the linear segments for the various ondition ratings end-to-end, as in Figure 3, will produe a deterioration urve indiating the relationship between the ondition rating and time. However, two problems are assoiated with the data and adjustments must be made. First, it was reognized that many bridges are either rehabilitated or replaed as the element ondition rating delines to lower levels. Thus, the number of bridges dropping to a lower ondition rating beomes small ompared with the number improving or remaining unhanged. This an make the numerator of Equation 4 very small ompared with the denominator. As a result, the average hange alulated would be very small. Hene, the time alulated to deline to a lower ondition rating would be overestimated. (4) The seond problem was a signifiant number of I-point inreases in ondition that were not learly linked to rehabilitation. After onsultation with bridge maintenane experts from NCDOT, it was onluded that the improvement of 1 point is sometimes the result of a different onlusion by a subsequent inspetor. This upgrading of ondition an our in borderline ases beause of the general nature of the ondition rating definitions. Another ause for the I-point improvements was attributed to the effets of very minor work or preventive maintenane. As a result of the first problem it was determined that onsidering the observations for improved bridges in the analysis was essential for finding a reasonable solution. A rational method of aounting for the deline before improvement was needed. The bridge element ondition rating at r (Figure 6) would have delined to a lower ondition rating _if the work had not been performed. However, it was not possible to determine how soon this would have ourred. As illustrated in Figure, the improvement might our immediately after a deline to r suh as / or at any later time up to just before the time that it would have delined to r - 1. Assuming a normal distribution of the improvement timing, a reasonable approximation would be to assume a timing as shown by 1 _ 5 This is equivalent to assuming that the rehabilitation is oinident with a deline of one-half point to r -.5. In reality, a ondition rating annot drop by half of a point. However, the number of observations that had improvement indiated from ondition rating r were assumed to deline by j =.5. As for problem two, it was determined on the basis of the advie of experts from NCDOT to exlude the data for whih there was a

14 TRANSPORTATION RESEARCH RECORD 149 O> r+1!ti r-.5 :. () r-1 Range of Improvement Times 1....I 11 lo.s 11 -- -- -J t..._.t ' - ',, i Average Duration at r 1,.... I FIGURE Timing of improvements. Time I-point improvement. Thus, the number of observations that had an improvement of l point was set equal to zero. Based on this, Equation 4 was reformulated to aount for these hanges. The general equation an be summarized as weighted no. of delines + 1/ no. of improving by> 1 point A VGCHN, = (5) total no. of bridges - no. of bridges improving by 1 point The new equation was thus L L n(r.t.j) X J + L L (n(r.t.j) X.5) YRL-l 111 YRL-l z t=yrl j= I t=yrl j=- YRL- I 111 YRL- I I I n(r.r.j) - I n(r.1.-i) t=yrl j=z t=yrl where z is the maximum number of points the bridge element an improve from r (i.e., 9 - r). This methodology was used to develop deterioration urves for the three major bridge elements. Data on bridge ondition ratings for the dek, superstruture, and substruture and other bridge harateristis were extrated from the Bridge History file for the years 198 through 1989. Eah bridge element was also initially subdivided by the element material type. The results generated are illustrated elsewhere (9). Further subgroupings were then onsidered for eah element. Dek Groupings for Deterioration Analysis One of the main auses of dek deterioration is reinforement orrosion ind.ued by deiing salt. It was therefore desired to inlude the effets of deiing salts on the deterioration rates of the dek (6) bridge element. However, bridges on whih salt is used are not speifially defined in the NCBI. An alternate approah was used on the basis of input reeived from NCDOT engineers indiating that salt use is roughly limited to federal aid bridges in NCDOT geographi Divisions 5 and through 14 [Table 1 (a)]. These divisions are loated in the Piedmont and western parts of the state (Figure 8), where ie and snow are more frequent than in the eastern region. All other bridges were defined as nonsalted bridges. A seond variation of this grouping was based on dividing the salt region into two parts, the far west part of the state, whih inluded Divisions 11, 13, and 14, as Salt Region 1, and Divisions 5, 1, and through 1 as Salt Region. Salt Region 1 was thus loated in the oldest part of the state. Both salt regions inluded only federal aid bridges. All other bridges in these divisions and all bridges in Divisions 1 through 4 and 6 were ategorized as nonsalt bridges. The results indiated that the differenes between Salt Regions 1 and were not very signifiant. However, the effet of the ombined salt regions was very obvious ompared with that of the nonsalt regions. The grouping with one salt region of federal aid bridges versus all other bridges as nonsalt was therefore seleted. The effet of ADT on bridge deterioration was then onsidered. This was do.ne by dividing the ADT ranges into six subgroups as shown in Table 1 (a). Although some of the results generated for some of the subgroupings were reasonable, the overall pattern did not fit the. experiene enountered in pratie. The majority of those that did not fit the pattern were based on a very limited number of observations, in partiular for the upper and lower ranges of ADT. Bridges on different highway lassifiations are sometimes built to different standards. Therefore, the effet of highway funtional lassifiation on the deterioration rates of bridge deks was also investigated. Another advantage of using the highway funtional lassifiation as a way of subgrouping the bridges is that in general

Abed-Al-Rahim andlohnston 15 TABLE 1 Dek Groupings for Deterioration Analysis (a) Preliminary Trial Groupings Salt Region Classifiation Dek Material Type Funtional Classifiation Average Daily Traffi (ADT) Final Groupings Salt Region Classifiation Dek Material Type Funtional Classifiation I. II. a) b) ) d) a) b) ) d) a) b) ) d) a) b) Categories within Group Define Federal Aid bridges in Divisions 5 and -14 as salted bridges, vs other bridges and divisions as non salted; or Divide Bridges into 3 subgroupings: a) Divisions 11, 13, 14 (Federal Aid salted vs. others nonsalted) b) Divisions 5,, 8, 9, 1, 1 (Federal Aid salted vs. others non-salted) ) Divisions 1,, 3, 4, 6 (all non-salted) Reinfored onrete Cored slab and preast onrete Timber and laminated timber Steel plank Interstate, Prinipal Arterial, and Minor Arterial Major Colletor Minor Colletor Loal (b) Final - 1-8 81-1 - 4 41-8 ADT => 81 Categories within Group Define Federal Aid bridges in Divisions 5 and -14 as salted bridges, vs other bridges and divisions as non salted. Reinfored onrete Cored slab and preast onrete Timber and laminated timber Steel plank Interstate, Prinipal Arterial, Minor Arterial, and Major Colletor Minor Colletor and Loal there is an approximate relationship between the traffi volume and the type of highway. Thus, the effet of ADT on deterioration rates would be roughly aounted for by onsidering the type of highway lassifiation. Bridges were divided into four subgroups of highway lassifiations as indiated in Table I (a). The results generated were promising. However, ertain subgroupings still suffered from the lak of a suffiient number of data points. There were almost no observations for the minor olletor and loal routes in the salt regions. In addition, very limited data existed for the timber and steel deks on Interstates and arterials. The data for the nonsalt region were further analyzed by ombining the Interstate and arterials subgroup with the major olletors. The minor olletor and loal routes were also ombined into one subgroup. Data in the salt region were analyzed as one group. Table I (b) shows the final groupings for the dek deterioration analysis. Superstruture Groupings for Deterioration Analysis The effet of salt on the bridge superstruture ondition rating was onsidered from two perspetives. The first was deiing salt, similar to the earlier approah for bridge deks. The other was to study the effet of seawater sine orrosion-related deterioration an our in any area that is exposed and within the reah of the seawater spray (JO). The effet of the deiing salts was first studied. The superstruture elements in the salt region, espeially prestressed and reinfored onrete, tended to deteriorate at a faster rate than those in the nonsalt region. However, the effet was not signifiant for the steel and timber superstrutures. The effet of the seawater on the superstruture deterioration rates was then studied. Bridges were divided into two groups: those bridges in oastal ounties (Figure 8) and over a waterway were lassified as marine environment; all other bridges were lassified

16 TRANSPORTATION RESEARCH RECORD 149 DIV.13 DIV. 11 Ol'J. 9 DIV. DIV. 5 DIV. 4' OIV. FIGURE 8 Divisions in North Carolina. as nonmarine environment. The deterioration rates for the marine environment were greater than those for the nonmarine environment. It was also evident that the effet of the seawater was more signifiant than the effet of the deiing salts. Thus, the effet of the. seawater was seleted for further analysis. Bridges were then subgrouped by funtional lassifiation, similar to the approah used for bridge deks. The superstruture type was another parameter thought to influene the deterioration rates of the bridge superstruture. Reinfored onrete and steel strutures were therefore divided into two subgroups eah, as shown in Table (a). The steel truss subgroup did not ontain suffiient numbers of observations when it was subdivided by highway lassifiation and marine versus nonmarine environment. Thus, all steel trusses were analyzed as one group. The differene in the deterioration rates for the onrete struture types was very small. The subgroups were therefore ombined together under reinfored onrete. However, steel trusses tended to deteriorate at a faster rate than the other types of steel strutures. The final groupings for the superstruture deterioration analysis are given in Table (b). Substruture Groupings for Deterioration Analysis The effet of seawater on the deterioration rates of substruture elements was first studied. Bridges were thus divided into groups of marine and nonmarine environments similar to the approah used for the superstruture analysis. However, the nonmarine environment group was further divided into waterway and grade separation. This was done. so that the effet of freshwater on the bridge substruture ould be evaluated. Bridges were also subgrouped by material type. DETAILED DETERIORATION RESULTS Detailed deterioration urves for bridge deks are plotted in Figure 9. From these urves it is apparent that the deiing salt aelerates the deterioration of bridge deks. The effet of the deiing salt was most signifiant on the prestressed onrete deks; this was followed by the reinfored onrete. The effet of the deiing salt on the steel (asphalt-filled steel pan) and timber deks was very small, as might be expeted. It an be noted that the prestressed onrete generally has higher ondition ratings at early ages, but one problems our the ondition ratings deline rapidly. This is probably due to -reognition by inspetors that evidene of a problem in prestressed members an be a major onern sine the small area of steel is sensitive to orrosion or other forms. of deterioration. Bridge deks loated on minor olletor and loal routes tended to deteriorate at a slower rate than those loated on Interstate, arterial, and major olletor routes. This ould be attributed to the higher volume of traffi and the higher perentage of truks that use the latter types of highways. Ptestressed onrete was the only exeption to this trend, possibly beause of variations in the design of prestressed onrete deks. As for the deterioration rates of the bridge superstruture element, it was evident tha( the salt from the sea air or water splash inreased the deterioration rates of the element. It was also evident that bridges on Interstate, arterial, and major olletor routes deteriorated at a faster rate. than those on minor olletor and loal routes. However, the differene in the. deterioration rates of the superstruture rates between the two types of highway groupings was not as signifiant as the differene in the deterioration rates of bridge deks. This ould be attributed to the fat that the impat of traffi on the superstruture is not as severe as that on deks. The deterioration urves generated for superstruture elements, subdivided by material and other groupings, an be found elsewhere (9). Bridge substrutures loated in a marine environment were found to deteriorate at a muh faster rate than those loated in a nonmarine environment. In addition, those bridges that were over a waterway tended to deteriorate at a faster rate than bridges at a grade separation but at a slower rate than the bridges in a marine environment. The deterioration urves generated for substruture elements, subdivided by material and other groupings, an be found in elsewhere (9).

Abed-Al-Rahim and Johnston 1 TABLE Superstruture Groupings for Deterioration Analysis (a) Preliminary Trial Groupings Categories within Group Marine a) Marine Environment: In a oastal ounty shown in Figure 8 and Environment over a waterway Classifiation b) Non-marine Environment: All other bridges not inluded in the marine environment ategory Salt Region Define Federal Aid bridges in Divisions 5 and -14 as salted Classifiation bridges, vs other bridges and divisions as non salted.. Material and a) Prestressed Conrete Struture Type b) Reinfored Conrete i) Slab and M-beam ii) T-beam, Girder Floor Beam, Box Beam (Multiple and Single) ) Steel i) Truss (Thru and Dek) ii) All other Types d) Timber Funtional a) Interstate, Prinipal Arterial, and Minor Anerial Classifiation b) Major Colletor ) Minor Colletor d) Loal Final Groupings (b) Final Categories within Group Marine a) Marine Environment: In a oastal ounty shown in Figure 8 Environment and over a waterway Classifiation b) Non-marine Environment: All other bridges not inluded in the marine environment ategory Material and a) Prestressed Conrete Struture Type b) Reinfored Conrete ) Steel i) Truss (Thru and Dek) ii) All other Types d) Timber Funtional a) Interstate, Prinipal Arterial, Minor Arterial, and Major Classifiation Colletor b) Minor Colletor and Loal Overall, the analysis produed a set of results that is onsistent with a rational omparative onsideration of the material, environment, and other fators. SUMMARY AND CONCLUSIONS A methodology was developed for prediting the deterioration rates of the bridge dek, superstruture, and substruture elements as measured by FHW A bridge inspetion ondition ratings. A set of deterioration urves was developed for the three major bridge elements by material type. Another set of deterioration urves was developed for various subgroupings of the bridge elements on the basis of environment and funtional lassifiations. 1. For deks deiing salts were found to ause the deterioration rates to inrease, in partiular for prestressed and reinfored on"' rete deks. The effet of the deiing salts on the timber deks was not very signifiant. The highway lassifiation was signifiant.in relation to the deterioration rates of the bridge deks. Bridge deks on minor olletor and loal routes tended to deteriorate at a slower rate than those on Interstate, arterial, and major olletor routes. This was attributed to the higher traffi volumes and the higher perentage of truks that use the latter type of highways.. Deteri6ration rates. for the superstruture tended to be higher for those bridges exposed to the splashing of saltwater than those that are not exposed to saltwater. Bridge superstrutures on Interstate, arterial, and major olletor routes were found to deteiiorate at a faster rate than those on minor olletor and loal routes. 3. The effet of saltwater was found to ause a rapid inrease in the deterioration of the bridge substruture. Although freshwater Was also found to inrease the deterioration rate of the substruture, the impat was not as signifiant as that.of saltwater. Substruture deterioration rates at grade separation were omparatively low.

18 TRANSPORTATION RESEARCH RECORD 149 Q 9 9 8 Non-Nit Region 8.An.W, Md Mlljar Calle:tor 6 8 Q i 5 ii 5 r: 4 4 :.;; :.ti: 1::1 1::1 3 3 1 3 4 5 8 1 3 4 5 8. llme (years) rune (Years) (a) (b) 9 8 6 6 C> C> i 5 i 5 4. 4 :iii Non-ult Region...: 1::1 Mnor Colletor and l.oll 1::1 3 3 9 8 Non-ult ReiJion Interstate, Arterial, 81111 Mai«Colleda' 1 3 4 5 6 1 3 4 5 6 rune (Years) Time (Years) () (d) FIGURE9 Dek ondition rating versus time: (a) prestressed onrete, (b) reinfored onrete, () steel, and (d) timber. ACKNOWLEDGMENTS The researh desribed herein was onduted at North Carolina State University with the support ofncdot. The writers gratefully aknowledge the support provided by NCDOT and the assistane provided by Jimmy D. Lee, State Bridge Maintenane Engineer. REFERENCES I. O'Connor, D. S., and W. A. Hyman. Bridge Management Systems. Report FHWA-DP-1-lR. FHWA, U.S. Department of Transportation, Ot. 1989.. Busa, G., M. Cassella, W. Gazda, and R. Horn. A National Bridge Deterioration Model. Report SS-4-U5-6. Transportation Systems Center, U.S. Department of Transportation, Cambridge, Mass., 1985. 3. Hyman, W., D. Hughes, and T. Dobson. The Least Cost Mix of Bridge Replaement and Repair Work on Wisonsin's State Highways Over Time-A Computer Simulation. Draft Tehnial Report. Wisonsin Department of Transportation, Madison, 1983. 4. Chen, C. J., and D. W. Johnston. Bridge Management Under a Level of Servie Conept Providing Optimum Improvement Ation, Time, and Budget Predition. Report FHW A/NC/88-4; Researh Projet 86-. Center for Transportation Studies, Department of Civil Engineering, North Carolina State University, Raleigh, 198. 5. Jiang, Y., and K. C. Sinha. Bridge Servie Life Predition Model Using the Markov Chain. In Transportation Researh Reord 13, TRB, National Researh Counil, Washington, D.C., 1989, pp. 4-3. 6. Saito, M., and K. C. Sinha. A Delphi Study in Bridge Condition Rating and the Effet of Bridge Improvement Alternatives. Presented at 69th Annual Meeting of the Transportation Researh Board, Washington, D.C., 199.. Reording and Coding Guide for the Struture Inventory and Appraisal of the Nation's Bridges. FHWA, U.S. Department of Transportation, De. 1988. 8. Harper, W. V., A. Al-Salloum, S. Al-Sayyari, S. Al-Theneyan, J. Lam, and C. Helm. Seletion of Ideal Maintenane Strategies in a Network Level Bridge Management System. In Transportation Researh Reord 168, TRB, National Researh Counil, Washington, D.C., 199, pp. 59-6. 9. Abed-Al-Rahim, I. J., and D. W. Johnston. Analysis of Relationships Affeting Bridge Deterioration and Improvement. Researh Projet 9-6. Center for Transportation Engineering Studies, Department of Civil Engineering, North Carolina State University, Raleigh, 1991. 1. Novokshheno, V. Salt Penetration and Corrosion in Prestressed Conrete Members. Publiation FHW A RD-88-68. FHW A, U.S. Department of Transportation, July 1989. The ontents of this paper reflet the views of the authors, who are responsible for the fats and auray of the data presented herein. The ontents do not neessarily reflet the offiial views or poliies. of NCDOT. This paper does not onstitute a standard, speifiation, or regulation. Publiation of this paper sponsored by Committee on Strutures Maintenane and Management.