# Condition assessment model for sewer pipelines using fuzzy-based evidential reasoning.

1. IntroductionThe condition and level of service of sewer pipelines could have major effect on the environmental and economic aspects for populated urban areas. Deteriorated sewer pipelines are considered hazardous on the environment and public health and safety. Condition assessment models for sewer pipelines can reflect their physical state, which may have an effect on the asset's performance. The performance of sewer pipelines is a function in reliability, and level of service by which decision-makers can determine the lifecycle of the pipeline and the time for interventions required to reinstate the level of service of deteriorated pipes back to the desired level. Maintenance, rehabilitation and renewal comprises the intervention of assets, which can be associated with knowing the current condition of the pipelines. In absence of information regarding the condition of pipelines, unforeseen failure can take place making asset replacement inevitable which is the most expensive measure amongst the rest intervention measures. The presence of condition assessment models can help in managing assets and avoiding early failure. It can also provide an accurate prediction of expenses required in the future through understanding and predicting the remaining asset life and its condition. Also, condition assessment models can help in avoiding erroneous observations of inspected sewer pipelines as a result of human mistakes and poor personal judgement of Closed-Circuit Television (CCTV) operators. Condition assessment models can help in better maintenance and rehabilitation strategies by determining the required corrective actions (i.e. maintenance, rehabilitation, renewal) and their timeframes. A reliable inspection plan and condition assessment models for the maintenance and rehabilitation of sewer pipelines is needed to control and minimise potential adverse effects of assets' failure (National Guide to Sustainable Municipal Infrastructure best practice, 2003).

Several researches have addressed the topic of modelling the condition of sewer pipelines. In the context of predicting the condition of sewer pipelines, a model was developed taking into consideration the pipe material, length, diameter, decreased flow capacity of the pipes, kind of roads above pipes, inflow and infiltration (Hasegawa, Wada, and Miura 1999). The developed model could estimate the condition of existing sewers by analysing their performance and calculating the change in flow in pipes, from which a decision can be made regarding how necessary repair is required. Although this was one of the very first attempts to model deterioration of pipelines, the developed model was not efficient in evaluating a sewer's conditions due to its complexity (Fenner 2000). Another model was developed for assessing the conditions of sewers using logistic regression technique (Ariaratnam, El-Assaly, and Yang 2001). Historical data were used to develop this model using age, size, material, effluent type and burial depth as affecting factors. It was found that some of the factors used were insignificant such as the depth and material type and that the increase in age would lead to an increase in the probability of deficiency. Although using logistic regression provided a flexible model in which qualitative and quantitative variables were used to determine sewer pipelines' criticality, the model was based on prior knowledge of historical inspection records which might not be practical in many cases. The effect of pipes' location, diameter, burial depth and their function were used in assessing the condition of large diameter sewer pipelines (Mcdonald and Zhao 2001). This research provided guidelines in which the condition and impact ratings were combined to prioritise inspection and rehabilitation while taking into consideration the impact of only six affecting factors, without presenting a condition assessment model. Transition curves were also used to develop methodologies for forecasting the condition of sewers (Baur and Herz 2002). Historical data for sewer conditions were used in developing these transition curves. The remaining life of the pipe was predicted by determining the ageing speed using transition curves. The generated transition curves allowed the user to identify pipe segments in each condition state, however cohort (i.e. the classes that sewer pipes share) of pipes had to be generated to create these transition curves which require a considerable amount of data. Condition rating of sewer pipelines was estimated through a model that was built using rule-based simulation (Ruwanpura, Ariaratnam, and El-Assaly 2003). The presented model deployed age, material type and length in which the deterioration pattern for points with no data were assumed to be the same as points with known data which made the model's accuracy questionable. Artificial Neural Networks (ANN) using historical data were used to develop a condition assessment model for sewer pipelines (Najafi and Kulandaivel 2005). The model took into consideration age, size, length, material, burial depth, slope and effluent type. It was found that the pipe's diameter was the most significant variable and slope was insignificant. In the development of this model, an extensive data-set was used to learn all possible combinations that can be considered one of the limitations of this model due to scarcity of data. Also, Tran, Perera, and Ng (2009), developed a structural prediction model for storm water pipelines using Neural Networks with the aid of Markov chain. Factors such as diameter, age, depth, slope, and number of tress, hydraulic condition, location of pipeline and soil type were considered in this model. By checking the performance of the developed model, it was found that the overall success rate was 81%. In another research, the effect of effluent type was studied on concrete sewer pipelines subjected to sulphide (Mahmoodian and Alani 2016). Condition assessment models using multiple regression were developed to predict the structural and operational condition ratings for sewer pipelines by Chughtai and Zayed (2007a, 2007b, 2008). Eight variables were included in the formulation of the structural condition equation which were: age, size, material type, material class, burial depth, length, bedding material and street type. As for the operational condition assessment model, only five factors were used in the equation formulation. In order to determine the significance of the predictor variables, a best subset analysis was performed. Although the validity of the proposed models was approximately 80%, one of the assumptions that were made to develop these models was assuming a linear relationship between the dependent and explanatory variables, which is not the case in most of the deterioration patterns in sewer pipelines. In an attempt to address uncertainties in assessing the conditions of infrastructure buried pipelines, a model using hierarchical evidential reasoning was proposed by Bai et al. (2008). This model considered only four factors related to the inner wall surface condition of the pipeline. The developed model could deal with missing data without making strong assumptions but didn't take into consideration the dependencies between contributing factors. Several research studies have addressed performance of sewer pipeline in context of reliability (Mahmoodian and Alani 2013a, 2013b; Uslu et al. 2015). Evidential reasoning approach has been used in various civil and infrastructure applications such as structural systems (Bae, Grandhi, and Canfield 2004; Sadiq, Najjaran, and Kleiner 2006; Bolar, Tesfamariam, and Sadiq 2013), environmental decision-making (Chang and Wright 1996) and risk management (Tesfamariam, Sadiq, and Najjaran 2010). Additionally, other engineering applications using evidential reasoning can be found in Sentz and Ferson (2002). To include uncertainties and vagueness in the evaluation process, researches used Fuzzy Set Theory (FST) in many engineering aspects and infrastructures area (Sadiq, Rajani, and Kleiner 2004; Kleiner, Sadiq, and Rajani 2004a, 2004b; Fares and Zayed 2010; Baah et al. 2015; Anbari, Tabesh, and Roozbahani 2017; Marzouk and Osama 2017).

The aim of the aforementioned developed models was to predict the condition of existing sewer pipelines, taking into consideration some of the factors that affect their conditions. Some factors were not included and very few of the developed condition assessment models took into consideration the interdependencies between these factors and the uncertainties in the severity of weights between them. One of the main drawbacks of the previously developed condition assessment models is their dependence on data availability. One of the main sources of gathering information about factors are data from CCTV inspection reports which could be either incomplete or ambiguous resulting in erroneous and uncertain models. Additionally, sometimes gathering such data are costly or could be hard. This has raised the issue of data availability and reliability. The proposed condition assessment models in this study were built based on the experience of specialists working in different fields in drainage networks to overcome the problem of data availability as the collected data-sets were only used for validation purposes. Therefore, the accuracy of the developed models were not impacted by the limited amount of data available where the data have only been used for validation purposes and yielded accurate predictions. It is anticipated that using a more comprehensive data-set which include more factors will enhance the validation outcome. In addition, this study provided the weight of the impact of each factor on the pipeline conditions to benefit practitioners in prioritising their data collection based on the importance of the identified factors. The intent of this paper is to develop a model that assesses the overall condition of gravity sewer pipelines. The model utilises FST, and Evidential Reasoning (ER) that incorporates Multi-Criteria Decision Analysis (MCDA) under uncertainties with the aid of Fuzzy Analytical Network Process (FANP) integrated with Monte-Carlo Simulation. This approach incorporates the interdependencies between contributing factors and uncertainties inherited when deriving their relative weights. The evidential reasoning was used as an aggregation method, using the concept of degree of assurance which is useful when there are insufficient information in the assessment process. Fuzzy Logic was combined with evidential reasoning to overcome the uncertainty that might accompany human judgments when using linguistic terms. Experts working in the field of infrastructures and wastewater collection networks were sought to determine the relative weights and effect values for the different factors affecting sewer pipelines through questionnaires. The developed model is validated using a real data-set from an existing sewage network in Doha in the State of Qatar.

2. Model development

The main idea to develop the proposed condition assessment model is to integrate the factors contributing to the deterioration of sewer pipelines and their effects to create an index representing the condition of pipelines. Figure 1 summarises the methodology adopted in this research. The first part of the methodology was identifying and collecting data related to the factors that would affect and contribute in the deterioration of sewer pipelines. After identifying these factors, experts working in the field of infrastructures and sewage networks were sought through a questionnaire to determine the importance of the contributing factors relative to each other and to the overall pipeline condition in addition to the severity of their effect on sewer pipeline condition. To overcome the uncertainty associated with human judgments in determining the relative weights of the contributing factors, FANP associated with Monte-Carlo Simulation were used to determine the final relative weight values. In the same manner, FST was used to assign the fuzzy membership functions and thresholds for the severity of the factors' effects on the pipeline conditions. In order to combine both the effect values and relative weights of factors affecting sewer pipelines, ER was used as an aggregation technique. It was used to determine the credence in the model's outputs that represents the user's certainty level about how good the condition of the pipeline is, which is based on the different contributing factors' effect values. After combining all the factors degrees of belief, defuzzification using FST to generate a final crisp condition index was carried out. Finally, the developed model was validated using actual data for an existing wastewater collection network in the state of Qatar.

2.1. Identifying factors affecting pipeline conditions

To identify the factors affecting sewer pipelines' condition, previous studies addressing these factors (Fenner 2000; Fenner, Sweeting, and Marriott 2000; Ariaratnam, El-Assaly, and Yang 2001; Davies et al. 2001; Baur and Herz 2002; Hahn et al. 2002; Micevski, Kuczera, and Coombes 2002; Miiller 2002; Baik, Jeong, and Abraham 2006; Tran 2007; Ana et al. 2009) were investigated. After identifying the factors, an unstructured questionnaire was constructed and experts' knowledge was sought regarding the different factors' effect on sewer pipelines' condition

The experts included engineers and managers working in the field of infrastructures operation and maintenance, sewage networks design engineers, sewage network operators and contractors working in inspection of sewage networks in Doha in the state of Qatar. The majority of these experts (75%) were professional and chartered engineers with a working experience ranging between 15 and 25 years in the infrastructures field. Forty per cent of the addressed experts were managers, 50% were intermediate level engineers and 10% were technicians working in the inspection activities for sewer pipelines. Tables 1 and 2 show the distributed questionnaire in which experts were requested to fill two kinds of tables having the different factors contributing to the deterioration of sewer pipelines under gravitational and pressurised flows. First, the questionnaire sought the experts' opinions on the importance and influence of each factor, then on the effect values of each factor on the pipeline condition. Experts were required to perform pairwise comparison between the relative importance of the different contributing factors using Saaty's nine-point scale (Saaty 1996). As for the effect values of the different factors on the pipelines' condition, a 0-10 scale with zero as minimum and 10 as maximum was used to reflect the expert's opinions on the effect of each factor on the condition of pipeline.

The affecting factors were categorised into three categories namely: physical, environmental and operational. A breakdown for the different factors included in the questionnaire with a brief description is shown in Table 3. In total, 17 factors were identified and considered for the sewer pipelines. The physical factors included pipeline physical characteristics and components, while operational factors were those factors related to the hydraulics and flow inside the pipeline. As for environmental factors, they were the factors related to the surroundings of the pipelines.

2.2. Determining relative weights ([[omega].sub.i]) for factors using FANP and Monte-Carlo Simulation

Analytical network process (ANP) is considered as a MCDA technique that takes into consideration inter-dependencies between decision alternatives. However, ANP neglects uncertainties of human judgement when evaluating the pairwise comparison between the different factors. As such, Fuzziness has been introduced to ANP in the form of FANP where uncertainties and vagueness accompanied to human judgments are considered while taking into consideration the interdependency between different factors. ANP was used to model a three-level network representing all contributing factors and sub-factors to determine how strongly each of them affect sewer pipeline conditions. Fuzzy Preference Programming method was used to conduct FANP (Zhou 2012). Relative weights were determined as a solution for a non-linear maximisation problem where, the constraints in this problem were the upper and lower fuzzy numbers and the global weights were the objective of the problem. Steps of implementing the FANP using Fuzzy Preference Programming can be summarised as below:

(1) Rating and developing a network hierarchy for different elements with different levels to perform a pairwise comparison according to Saaty's scale.

(2) Developing a paired comparison matrices after comparing all elements using the fuzzifying scale. The Triangular Fuzzy Numbers (TFNs) shown in Table 4 were used in the fuzzification process in which crisp values derived from the pairwise comparison are transformed into fuzzy inputs by creating a lower, upper and most likely matrices.

Using Fuzzy Preference Programming (FPP) as per Equation (1) to defuzzify the developed lower, upper and most likely matrices to calculate the global relative weights (Zhou 2012). The FPP provides a solution to a prioritisation (optimisation) problem with (n) elements, in which (m) fuzzy pairwise comparisons are represented by triangular fuzzy numbers [??]. The goal of the FPP problem is to determine a priority vector ([[omega].sub.k) in a way that ([W.sub.i]/[W.sub.j]) known as the priority ratio value would be in between the initial fuzzy pairwise comparison values.

[mathematical expression not reproducible] (1)

(1) where [min.sub.ij], [mid.sub.ij], [max.sub.ij], are minimum, middle and maximum limits of triangular fuzzy number, while ([lambda]) and ([[omega].sub.k) represents the solution of the formulated optimisation problem. The value of (A) measures the consistency of fuzzy judgement of the FANP problem, with a maximum value of 1, whereas [[omega].sub.k] is priority crisp vector representing the relative weights of the factors, [[omega].sub.k] > 0, k = 1,2,3,...,n, i = 1,2,3,...,n-1 and j = 2,3,...,n, j > i

(3) Constructing an un-weighted super-matrix, using the relative weights of the obtained defuzzfied factors.

(4) Obtaining weighted super-matrix in which columns in the unweighted supermatrix are normalised individually by dividing each element by the summation of the column elements.

(5) Calculating the limit matrix by raising the weighted super-matrix to large powers until the convergence is achieved. The limit matrix is the sum of powers by which the super-matrix was raised until all its entries become zero. Table 5 shows a sample for the un-weighted, weighted and limit matrices.

(6) Determining relative weights for different factors from the resulting limit matrix.

Based on the collected 36 questionnaires' (i.e. responses), some of the factors had different weights. Hence, probability distributions were developed using the factors' different weights in Monte-Carlo Simulation (MCS).

MCS consists basically of two operations; the first one is sampling in which random values are generated based on a certain probabilistic distribution and are considered the inputs. Then in the second operation, several iterations are carried out and samples are driven from each iteration. At the last iteration, a single output value is generated representing the output distribution allowing the user to identify the probability distribution associated with the output. This simulation technique was used to eliminate uncertainties inherited as a result of different weights from different questionnaires. Figure 2 shows the final weights calculated from the mean of the generated probabilistic curves resulting from MCS.

2.3. Effect values membership functions for factors using FST

FST (Zadeh 1965) is a mathematical model which is considered as a generalisation for classical (or crisp) sets concepts. FST attempts to provide a better tool to deal with vague situations that cannot be captured by the conventional crisp sets. In the crisp sets, functions membership is considered as a binary code, where the function membership is equal to 1 if the element belongs to the set and is 0 if the element doesn't belong to the function. The generalisation made by Zadeh (1965) was to produce Fuzzy sets that allows the membership function to have a gradual transition with any degree of membership from none to full. Unlike crisp variables that ignore uncertainty when dealing with them, fuzzy variables are capable of expressing uncertainties. Fuzzy numbers 'F' can be represented by a set [[a.sub.1], [a.sub.2], [a.sub.3], [a.sub.4]] which is defined by a membership function [[mu].sub.A] and can be expressed by Equation (2).

[mathematical expression not reproducible] (2)

Membership Functions (MFs) are the building blocks of FST, where they represent the fuzziness in a fuzzy set. Accordingly, the shapes of MFs are important for each particular problem. MFs may have different shapes like triangular, trapezoidal, Gaussian and others and can vary between 0 and 1.

To determine the inputs to be used in the ER module, linguistic factors' fuzzy thresholds for the effect values and their corresponding membership functions were generated using the questionnaires' responses. The methodology of implementing the FST starts with defining the limits for each linguistic factor affecting the pipeline conditions based on the questionnaires' responses. To represent the output for the effect of the different linguistic factors, a five-grade fuzzy subset is used {Excellent, very good, good, fair and critical}. After determining the input and output thresholds values, trapezoidal curves at the boundaries and triangular curves for the intermediate points were chosen as a membership function shapes to represent the inputs and output as shown in Figures 3 and 4. Triangular and trapezoidal fuzzy membership function shapes were used because they are suitable for representing linguistic variables (Lee 1996). Membership functions have been divided into four groups to appropriately take into account the age effect on the developed model. In such way, better and reliable assessment for new pipelines is assigned over the older ones by taking into consideration the age influence on the pipeline's deterioration. Figures 3-6 show the plot of the generated membership functions for diameter, length and the buried depth for gravity pipelines and their corresponding shapes for the fuzzy thresholds. Similarly, membership functions were generated for the rest of the factors affecting gravity and pressurised pipelines. The figures were developed by fuzzifying the input to determine the MF, for which each factor belongs. The corresponding MF ([[mu].sub.F](x)) based on he five-grade scale was calculated using Equation (2). The overlap between thresholds for the different membership functions at some intervals is due to the uncertainty in determining the cut-off values of the factors. Also, the trapezoidal shapes represent the extreme limits of excellent or critical membership functions for the thresholds, whereas triangular shapes are used to represent the three remaining membership functions in between the trapezoidal shapes.

2.4. Condition assessment index for sewer pipelines using ER

ER offers a rational methodology to deal with uncertainty, incompleteness and fuzziness for data aggregation which was concluded by Dempster-Shafer (D-S) theory of evidence (Yang and Sen 1994; Yang and Singh 1994; Yang 2001). ER utilises the concept of degree of beliefs which is useful in handling various types of uncertainties, in which each attribute of an alternative is represented by a distributed assessment using a belief structure in the MCDA problem. Unlike conventional approaches that require scaling grades and averaging scores to aggregate attributes, the ER aggregates degrees of belief approach by employing an evidential reasoning algorithm. In other words, ER allows aggregating different evidences by aggregating each two evidence and then aggregating the resultant with the third evidence and so on (Yang and Xu 2002).

In this research, the ER approach was used to determine the final condition index for sewer pipelines. The first step in implementing the ER module was to identify the evaluation grades (E), express them in a linguistic terms (i.e. Excellent, very good, good, fair and critical) and to state the final weight ([[omega].sug.i]) for each contributing factor. The belief structure of the formulated ER problem consisted of degree of beliefs indicating the user's level of certainty about the condition of the pipeline (Excellent, very good, ..., etc.) based on the different contributing factors effect values. For instance, the degree of belief is said to be high for an excellent pipeline condition (new pipe) with low-effect values of the contributing factors, while the degree of belief is considered low for very good pipeline conditions assuming an old pipeline with higher effect values for the contributing factors. After identifying the evaluation grades and relative weights, degrees of user's certainty (i.e. degree of belief) were transformed into basic probability masses ([p.sub.n;i]) using Equation (3), which calculates the product of degrees of belief and relative weights.

[P.sub.n,i] = [P.sub.i] ([E.sub.n]) = [[omega].sub.i][[beta].sub.n,i] n = 1, ...,N;i = 1, ..., I (3)

where p is the degree to which attribute (i) would result in a true hypothesis of attribute (y) (i.e. pipeline condition) until the (nth.) grade has been assessed. To determine ([p.sub.E,i]) which represents the unassigned probability masses remaining after assessing all (N) grades, Equations (4)-(6) were used.

[mathematical expression not reproducible] (4)

[mathematical expression not reproducible] (5)

[mathematical expression not reproducible] (6)

where [??] and [??].is the unassigned probability masses remaining after assessing all (N) grades. While, [??] is the probability mass with no evaluation grades resulting from missing or incomplete evaluation.

To calculate [??] the basic probability masses [p.sub.n,j] and [p.sub.E,j] for n = 1, ..., N and J = 1, ..., i were aggregated using Equation (9) which combines the two probability masses using normalising factor [F.sub.I(i+1)].

The normalising factor can be defined as [??].

where [F.sub.I(i+1)] is the normalising factor such that, summation of [p.sub.n,I(i+1]) + [p.sub.E,I(i+1]) for n = 1, ..., N is equal to unity.

[mathematical expression not reproducible] (7)

[mathematical expression not reproducible] (8)

[mathematical expression not reproducible] (9)

By combining each two probability masses until all the factors were combined, using Equations (10) and (11), the final probability masses and final degrees of beliefs were computed.

[mathematical expression not reproducible] (10)

[mathematical expression not reproducible] (11)

[mathematical expression not reproducible] (12)

where [[beta].sub.n] is the final degrees of beliefs for all assessed grades ([E.sub.n]), while ([[beta].sub.H]) represents the unassessed degrees of beliefs associated to (E). Defuzzificaion was carried out for the aggregated final assessment resulting from the ER module by utilising the FST module to determine the overall condition index for sewer pipelines. The defuzzification process is defined as calculating the areas of the resulting figures for each MF, where the weighted average method was used to convert the final degrees of belief into crisp values.

3. Model validation

To validate the developed sewer pipeline condition assessment model presented in this paper, actual conditions of existing pipelines in Qatar were compared with the results from the developed model. The actual pipelines condition data-set was obtained from the Public Work Authority located in Doha, Qatar. The actual condition assessment scores were extracted from CCTV analyses based on the sewer condition assessment code and class method as per the British (BSI 2012) and German (DWA 2015) standards, respectively.

The obtained validation set comprised of 549 pipelines with information such as size, length, age, groundwater table level and depth as the contributing factors and the pipeline's overall condition grade. The developed model was adjusted accordingly to include the validation set factors only. The actual condition assessment was evaluated based on 1-5 scale, with 1 indicating that the pipeline is in its best condition and no maintenance is required, while 5 indicates the worst condition for the pipeline and maintenance is required immediately. The predicted mean condition index values for the condition index of each pipeline were adjusted by converting the ranges of 0-2, 2-4, 4-6, 6-8, 8-10 to 5, 4, 3, 2 and 1, respectively, using rounding up and down for the number of the first decimal point (i.e. 8.2 rounded down to 8 and 8.8 rounded up to 9).

To examine the accuracy of the developed model the relationships recommended in literature by Zayed and Halpin (2005), Al-Barqawi and Zayed (2006) and Oberkampf and Roy (2010) were used. In order to determine the validity of the models, Equations (13) and (14) were used. When Average Validity Percentage (AVP) values are closer to 100, and Average Invalidity Percentage (AIP) values are closer to 0, this means that the model is valid with an average validity value of the AVP and invalidity value of AIR Equations (15) and (16) were also used to determine the accuracy of the model's prediction. As for Equation (17), it was used to measure the goodness of the developed model fitness to real data. The developed condition assessment model 549 pipelines resulted in 82%, 18%, 0.14, 0.19 and 875 for AVP, AIP, Root-Mean-Square Error (RMSE), Mean Absolute Error (MAE) and [f.sub.i], respectively. Table 6 shows a sample for the actual condition versus the model's output and the relevant information of the validation data-set used.

[mathematical expression not reproducible] (13)

AVP =100-AIP (14)

[mathematical expression not reproducible] (15)

[mathematical expression not reproducible] (16)

[f.sub.i] [[1000]/[1 + MAE]] (17)

where [E.sub.i] = predicted value, [C.sub.i] = actual value, n = number of observations and [f.sub.i]] = fitness function.

4. Conclusions

The development of sewer pipelines condition assessment model using FANP integrated with MCS, FST and ER techniques was presented in this paper. The proposed model incorporates interdependencies and MCDA under uncertainties. Seventeen factors were included in the model as the factors that affect the conditions of sewer pipelines. Experts working in the field of infrastructures and sewage networks were sought to determine the importance and degree of influence of the different factors through unstructured questionnaires. The considered factors were grouped under physical, environmental and operational categories. The degree of influence for each factor and category was determined by the FANP and Monte-Carlo Simulation module that considered the uncertainties and fuzziness associated with human judgments. To create membership functions and thresholds for the effect values of the different affecting factors, FST module was used. The overall condition index was determined using the ER module with the aid of FST in which degrees of belief were set and combined with different relative weights of the different factors. The FST was used to deffuzify the Fuzzy membership functions resulting in crisp values from the fuzzy overall condition. In order to validate the proposed model actual inspection data for 549 existing sewer gravity pipelines in Qatar were used. The validation results were 82%, 18%, 0.14, 0.19 and 875 for AVP, AIP, RMSE, MAE and [f.sub.i], respectively, which indicates that the developed model would yield in sound and reliable results. The proposed model can be used to get an overview about the current and future conditions of the existing pipes to develop proper plans for inspection, set up rehabilitation strategies and avoid catastrophic failures. In addition, this study provided the weight of the impact of each factor on the pipeline conditions to benefit practitioners in prioritising their data collection based on the importance of the identified factors. The proposed condition assessment model could be considered as a tool that would allow key personnel to properly plan their inspections, collect only necessary data and provide cost-effective rehabilitation and maintenance action based on the overall condition of sewer pipelines. Increasing the factors and number of pipelines included in model validation could improve the predictive accuracy of the proposed model. As such, it is recommended to calibrate the developed model using additional data-set that include additional factors.

Acknowledgement

The statements made herein are solely the responsibility of the authors. Also the authors would like to thank the public works authority of Qatar (ASHGAL) for their support in the data collection.

Disclosure statement

No potential conflict of interest was reported by the authors.

Funding

This publication was made possible by NPRP [grant number NPRP6-357-2-150] from the Qatar National Research Fund (a member of The Qatar Foundation).

Notes on contributors

Alaa Hawari is an associate professor in Qatar University and his research interests include water and wastewater treatment (membrane fouling, biosorption and selective adsorption) and water and sewer networks optimization.

Torek Zayed has PhD, MSc, and BSc in Construction Engineering and Management. He has more than 28 years of professional experience working in the construction industry training and in academic posts in USA, Canada and abroad. His research interests include infrastructures and assets management.

Mohamed Elmasry is a PHD candidate with more than 10 years working experience in the field of civil engineering. His research interests include construction and asset management.

Firas Alkadour is a civil engineer with more than four years of experience as a structural engineer. He holds a masters of science degree from the Department of Civil and Architectural Engineering at Qatar university.

References

Al-Barqawi, H., and T. Zayed. 2006. "Assessment Model of Water Main Conditions." Proc, Pipeline Division Specialty Conference. Chicago, IL: ASCE.

Ana, E., W. Bauwens, M. Pessemier, C. Thoeye, S. Smolders, I. Boonen, and G. De Gueldre. 2009. "An Investigation of the Factors Influencing Sewer Structural Deterioration." Urban Water Journal 6 (4): 303-312.

Anbari, M. J., M. Tabesh, and A. Roozbahani. 2017. "Risk Assessment Model to Prioritize Sewer Pipes Inspection in Wastewater Collection Networks." Journal of Environmental Management 190:91-101.

Ariaratnam, S. T., A. El-Assaly, and Y. Yang. 2001. "Assessment of Infrastructure Inspection Needs Using Logistic Models." Journal of Infrastructure Systems 7 (4): 160-165.

Baah, K., B. Dubey, R. Harvey, and E. McBean. 2015. "A Risk-based Approach to Sanitary Sewer Pipe Asset Management." Science of the Total Environment 505:1011-1017.

Bae, H., R. V. Grandhi, and R. A. Canfleld. 2004. "Epistemic Uncertainty Quantification Techniques including Evidence Theory for Large-Scale Structures." Advances in Probabilistic Mechanics and Structural Reliability 82 (13-14): 1101-1112.

Baik, H. S., H. S. Jeong, and D. M. Abraham. 2006. "Estimating Transition Probabilities in Markov Chain-based Deterioration Models for Management of Wastewater Systems." Journal of Water Resources Planning and Management 132 (1): 15-24.

Bai, H., R. Sadiq, H. Najjaran, and B. Rajani. 2008. "Condition Assessment of Buried Pipes Using Hierarchical Evidential Reasoning Model." Journal of Computing in Civil Engineering 22 (2): 114-122.

Baur, R., and R. Herz. 2002. "Selective Inspection Planning with Ageing Forecast for Sewer Types." Water Science and Technology 46 (6-7): 389-396.

Bolar, A., S. Tesfamariam, and R. Sadiq. 2013. "Condition Assessment for Bridges: A Hierarchical Evidential Reasoning (HER) Framework." Structure and Infrastructure Engineering 9 (7): 648-666.

BSI (British Standards Institution). 2012. Investigation and Assessment of Drain and Sewer Systems outside Buildings-General Requirements. London: BSI. ISBN: 978-0-580-70190-0.

Chang, Y. C., and J. R. Wright. 1996. "Evidential Reasoning for Assessing Environmental Impact." Civil Engineering Systems 14 (1): 55-78.

Chughtai, F., and T. Zayed. 2007a. "Sewer Pipeline Operational Condition Prediction Using Multiple Regression." Proc, Pipelines 2007: Advances and Experiences with Trenchless Pipeline Projects. Reston, VA: ASCE.

Chughtai, F., and T. Zayed. 2007b. "Structural Condition Models for Sewer Pipeline." Proc, Pipelines 2007: Advances and Experiences with Trenchless Pipeline Projects. Reston, VA: ASCE.

Chughtai, F., and T. Zayed. 2008. "Infrastructure Condition Prediction Models for Sustainable Sewer Pipelines." Journal of Performance of Constructed Facilities 22 (5): 333-341.

Davies, J. P., B. A. Clarke, J. T. Whiter, R. J. Cunningham, and A. Leidi. 2001. "The Structural Condition of Rigid Sewer Pipes: A Statistical Investigation." Urban Water 3:277-286.

Fares, H., and T. Zayed. 2010. "Hierarchical Fuzzy Expert System for Risk of Failure of Water Mains." Journal of Pipeline Systems Engineering and Practice 1 (1): 53-62.

Fenner, R. A. 2000. Approaches to Sewer Maintenance: A Review." Urban Water 2:343-356.

Fenner, R. A., L. Sweeting, and M. J. Marriott. 2000. "A New Approach for Directing Proactive Sewer Maintenance." Proceedings of the ICE-Water and Maritime Engineering 142 (2): 67-77.

German Association for Water, Wastewater and Waste (DWA). 2015. State Detection and Assessment of Drain and Sewer Systems outside Buildings-Part 3: Assessment by Optical Inspection. Hennef: DWA. ISBN: 978-3-88721-224-7.

Hahn, M. A., R. N. Palmer, M. S. Merrill, and B. Lukas. 2002. "Expert System for Prioritizing the Inspection of Sewers: Knowledge Base Formulation and Evaluation." Journal of Water Resources Planning and Management 128 (2): 121-129.

Hasegawa, K.., Y. Wada, and H. Miura. 1999. "New Assessment Systemfor Premeditated Management and Maintenance of Sewer Pipenetworks." Proc, 8th Int. Conf. on Urban Storm Drainage, 586-593. Sydney, Australia.

Kleiner, Y., R. Sadiq, and B. B. Rajani. (2004a). "Modeling the Deterioration of Buried Infrastructure as a Fuzzy Markov Process." Submitted to Journal of Infrastructure Systems, ASCE.

Kleiner, Y., R. Sadiq, and B. B. Rajani. (2004b), "Modeling Failure Risk in Buried Pipes Using Fuzzy Markov Deterioration Process." ASCE International Conference on Pipeline Engineering and Construction, San Diego, CA, August.

Lee, H. 1996. "Applying Fuzzy Set Theory to Evaluate the Rate of Aggregative Risk in Software Development." Fuzzy Sets and Systems 79: 323-336.

Mahmoodian, M., and A. M. Alani. 2013. "Multi-failure Mode Assessment of Buried Concrete Pipes Subjected to Time-dependent Deterioration, Using System Reliability Analysis." Journal of Failure Analysis and Prevention 13 (5): 634-642.

Mahmoodian, M., and A. Alani. 2013. "Modeling Deterioration in Concrete Pipes as a Stochastic Gamma Process for Time-dependent Reliability Analysis." Journal of Pipeline Systems Engineering and Practice 5(1): 04013008.

Mahmoodian, M., and A. Alani. 2016. "Sensitivity Analysis for Failure Assessment of Concrete Pipes Subjected to Sulphide Corrosion." Urban Water Journal 13 (6): 637-643. Marzouk, M., and A. Osama. 2017. "Fuzzy-based Methodology for Integrated Infrastructure Asset Management." International Journal of Computational Intelligence Systems 10 (1): 745-759.

Mcdonald, S. E., and J. Q. Zhao. 2001. "Condition Assessment and Rehabilitation of Large Sewers." Proceedings of International Conference on Underground Infrastructure Research, 361-369. Ontario.

Micevski, T., G. Kuczera, and P. Coombes. 2002. "Markov Model for Storm Water Pipe Deterioration." Journal of Infrastructure Systems 8 (2): 49-56.

Muller, K. 2002. "Development of a Generally Applicable Methodology for Selective Inspection of Sewer Networks. Part a: Scientific Studies." Final Report of the Institute of Urban Water Management ISA. Aachen: RWTH Aachen University.

Najafi, M., and G. Kulandaivel. 2005. "Pipeline Condition Prediction Using Neural Network Models." Proc, Pipeline 2005, 767-781. Reston, VA: ASCE.

Oberkampf, W. L., and C. J. Roy. 2010. Verification and Validation in Scientific Computing. Cambridge: Cambridge University Press.

Ruwanpura, J., S. T. Ariaratnam, and A. El-Assaly. 2003. "A Rule-based Simulation Models for Sewer Infrastructure." Proceedings of the Construction Research Congress. Honolulu, HI: ASCE.

Saaty, T. J. 1996. Decision Making in Complex Environments. the Analytical Hierarchy Process for Decision Making with Dependence and Dependence and Feedback. Pittsburgh, PA: RWS Publications.

Sadiq, R., B. Rajani, and Y. Kleiner. 2004. "Fuzzy-based Method to Evaluate Soil Corrosivity for Prediction of Water Main Deterioration." Journal of Infrastructure Systems 10: 149-156.

Sentz, K.., and F. Scott. 2002. "Combination of evidence in Dempster-Shafer theory." Vol. 4015. Albuquerque: Sandia National Laboratories.

Sadiq, R., H. Najjaran, and Y. Kleiner. 2006. "Investigating Evidential Reasoning for the Interpretation of Microbial Water Quality in a Distribution Network." Stochastic Environmental Research and Risk Assessment 21 (l):63-73.

Tran, H. D. 2007. "Investigation of Deterioration Models for Stormwater Pipe Systems." PhD thesis, Victoria University, Melburne, Australia.

Tran, H. D., B. J. C. Perera, and A. W M. Ng. 2009. "Markov and Neural Network Models for Prediction of Structural Deterioration of Storm-water Pipe Assets." Journal of Infrastructure Systems 16 (2): 167-171.

Tesfamariam, S., R. Sadiq, and H. Najjaran. 2010. "Decision Making under Uncertainty-an Example for Seismic Risk Management." Risk Analysis 30 (1): 78-94.

Uslu, B., S. K. Sinha, W L. Graf, and T. Angkasuwansiri. 2015. "Development of a Wastewater Pipeline Performance Prediction Model." In Pipelines 2015, 776-789. Baltimore, MD:ASCE VA.

Yang, J. B. 2001. "Rule and Utility Based Evidential Reasoning Approach for Multi-Attribute Decision Analysis under Uncertainties." European Journal of Operational Research 131:31-61.

Yang, J. B., and P. Sen. 1994. "A General Multi-Level Evaluation Process for Hybrid MADM with Uncertainty." IEEE Transactions on Systems, Man, and Cybernetics 24 (10): 1458-1473.

Yang, J. B., and M. G. Singh. 1994. "An Evidential Reasoning Approach for Multiple Attribute Decision Making with Uncertainty." IEEE Transactions on Systems, Man, and Cybernetics 24 (1): 1-18.

Yang, J. B., and D. L. Xu. 2002. "On the Evidential Reasoning Algorithm for Multiple Attribute Decision Analysis under Uncertainty." IEEE Transaction on Systems, Man, and Cybernetics, Part A: Systems and Humans 32 (3): 289-304.

Zadeh, L. A. 1965. "Fuzzy Sets." Information and Control 8: 338-353.

Zayed, T., and D. Halpin. 2005. "Deterministic Models for Assessing Productivity and Cost of Bored Piles." Journal of Construction Management and Economics 23 (5): 531-543.

Zhou, Xiaoguang. 2012. "Fuzzy Analytical Network Process Implementation with Matlab, MATLAB - A Fundamental Tool for Scientific Computing and Engineering Applications." Vol. 3, Prof. VasiliosKatsikis (Ed.), London: InTech. ISBN: 978-953-51-0752-1.

Alaa Hawari (a), Firas Alkadour (a), Mohamed Elmasry (b) and Tarek Zayed (b)

(a) Department of Civil and Architectural Engineering, Qatar University, Doha, Qatar; (b) Department of Building, Civil and Environmental Engineering, Concordia University, Montreal, Canada

CONTACT Alaa Hawaii a.hawari@qu.edu.qa

ARTICLE HISTORY

Received 29 March 2016

Accepted 8 January 2018

https://doi.org/10.1080/14488353.2018.1444333

Table 1. Description of the main and sub-factors affecting sewage pipelines conditions. Main factors Sub-factors Physical (PF) Age (AG) Diameter (Dl) Length (LE) Buried depth (D) Material (MT) Coating conditions (CQ Installation quality (IQ) Operational (OF) Flow rate (FR) Blockages (B) Infiltration and inflow (II) Corrosive impurities (CI) Maintenance and break strategies (MS) Operating pressure (OP) Environmental (EF) Soil type (ST) Bedding conditions (BC) Location (LO) Groundwater level (GW) Ground disturbance (GD) Main factors Description Physical (PF) Effects of pipeline degradation become more significant overtime The larger the pipe line diameter, the larger is its thickness, the lower is its deterioration rate and vice versa Longer pipes are more likely to suffer from bending stresses Life loads impact increases at shallow depths and the soil overburden impact increases at high depths. Moderate depths increase the life of sewers Different pipeline material show different failure patterns Pipelines with good coating conditions have higher resistance against corrosion Pipeline installation should be done according to certain standards and qualifications. High deterioration rates result from poor installation quality Operational (OF) High velocity water corrodes the internal walls of the pipe and causes disturbances especially when moving between pipes of different diameters. Low velocity water causes deposition and accumulation of sediments Accumulation of deposits and sediments, intrusion of trees roots and other types of blockages have a significant effect on the structural and operational condition of a sewer pipeline Infiltration washes soil particles and reduces the support along a pipeline Chemicals and substances (such as salts and micro-bio species) present in water while being transported decrease water quality. They can also corrode the internal surface of the pipe and cause its breakage The service life of sewer pipelines is increased by a good maintenance and break strategies The pressure resulting from transients in the water distribution system may cause pump and device failure, system fatigue or pipe ruptures Environmental (EF) This factor refers to the type of soil in direct contact with the pipe surface. Soils have different physical and chemical properties which have different impacts on the pipeline. Some soils are corrosive; others experience significant volume changes in response to moisture changes, resulting in changes to pipe loading. Presence of hydrocarbons and solvents in soil may cause pipe deterioration Sewer pipeline failure chances increases with improper bedding conditions Pipeline location is related to the installation zone being residential, industrial, school, etc Pipelines in residential areas are exposed to different conditions than those located in industrial areas or cities because they maybe located under different surface types (e.g. asphalt, seal, unpaved). City pipelines are subject to heavy traffic adding more dynamic load on the pipeline The amount of water in soil affects the soil resistivity, which inversely relates to the corrosion rate. The ground water may lead to corroding the pipe directly when salts and some corrosive substances exist in it Pipelines existing near a disturbed ground are subjected to high stresses and might have a sudden collapse Table 2. Questionnaire Sample for relative importance used in pairwise comparison for main factors and sub-factors in gravity and pressurised pipelines. Degree of importance Criterion (9) Absolute (7) Very (X) Strong (5) Strong 1. Main factors with respect to sewer pipeline condition Sewer pipelines condition Physical factors 2. Sub-factors with respect to each other Physical factors Pipeline age [check] Operational factors Corrosive impurities Environmental factors Groundwater level 3. Main factors with each other Physical factors Environmental factors Environmental factors Physical factors Operational factors Physical factors Degree of importance Criterion (3) Moderate (3) Moderate (5) (7) Very (X) (1) Equal Strong Strong Physical factors [check] [check] Pipeline age [check] [check] [check] Corrosive [check] [check] impurities [check] [check] [check] Groundwater level [check] [check] [check] Environmental factors [check] Physical factors [check] Physical [check] factors Criterion (9) Absolute (X) Criterion (y) Physical Environmental factors factors Operational factors Pipeline Pipeline diameter age Pipeline length Pipeline buried depth Pipeline material Pipeline coating conditions [check] Installation quality Corrosive Blockages (Ex. impurities Roots, Sediments) Infiltration and inflow Flow rate Maintenance and break record Groundwater Soil type level [check] Bedding conditions Location (Ex. Traffic Load) Ground disturbance (Ex. Construct on Work) Environmental Operational factors factors Physical Operational factors factors Physical Environmental factors factors Table 3. Questionnaire sample for effect values of different factors for gravity and pressurised pipelines. Qualitative Main Factor Sub-factors Unit of measure description (Parameters) Physical Pipeline age (Years) Old (>30) Medium (15-30) New(< 15) Pipeline diameter (m) Small (< 300) Medium (300-600) Large (>600) Installation quality (%) Poor Fair Good Operational Flow rate (m:7d) Low Medium High Maintenance and break (%) Poor strategies Fair Good Environmental Soil Type Rock Sand Ground disturbance (%) Low Moderate High Effect value on sewer Effect value on sewer pressurised pipelines Main Factor gravity pipelines (0-10) (0-10) Physical 8 9 6 8 4 6 8 5 6 5 4 6 10 10 7 8 4 5 Operational 2 3 3 4 6 7 8 10 5 7 1 3 Environmental 3 3 5 5 3 3 5 5 8 8 Table 4. Lower, most probable and upper limit matrices sample for operational factors in gravity pipelines. Lower limit matrix (*) Most probable matrix (*) Factors FR B II CI MS FR B II CI MS FR 1 41/2 41/2 2/15 2/15 1 5 5 1/7 1/7 B 2/11 1 1 1/9 1/9 1/5 1 1 1/9 1/9 II 2/11 1 1 1/9 1/9 1/5 1 1 1/9 1/9 CI 61/2 81/2 81/2 1 1 7 9 9 1 1 MS 61/2 81/2 1/9 1 1 7 9 1/9 1 1 Lower limit matrix (*) Factors FR B II CI CI FR 1 51/2 51/2 2/13 2/13 B 2/9 1 1 2/17 2/17 II 2/9 1 1 2/17 2/17 CI 71/2 9 9 1 1 MS 71/2 9 2/17 1 1 (*) Lower, Most probable and upper limit matrices values are as per the adjusted Triangular Fuzzy Number (TFN) matrix:; [??] Table 5. Unweighted, weighted and limit super-matrices sample for gravity pipelines. Un-weighted super-matrix Weighted super matrix Factors NC PF ... GD NC PF ... GD NC 0 0 ... 0 0 0 ... 0 PF 0.333 0 ... 0 0.333 0 ... 0 OF 0.333 0.167 ... 0 0.333 0.083 ... 0 ... ... ... ... ... ... ... ... ... SR 0 0 ... 0 0 0 ... 0 LO 0 0 ... 0 0 0 ... 0 GD 0 0 ... 1 0 0 ... 1 Limit super matrix Factors NC PF ... GD NC 0 0 ... 0 PF 0 0 ... 0 OF 0 0 ... 0 ... ... ... ... SR 0.019 0.013 ... 0 LO 0.169 0.121 ... 0 GD 0.169 0.121 ... 1 Table 6. Sample for the actual versus the resulting output of the model for the validation data-set. Actual data from the validation set Buried depth Pipeline (No.) Age Diameter (m) Length (m) (m) Water table (years) 1 17 200 17 2.40 Below pipe 2 17 200 36 1.90 Below pipe 3 17 200 41 1.30 Below pipe 4 17 200 25 4.60 Above pipe [??] [??] [??] [??] [??] [??] 465 33 150 49 3.2 Below pipe 466 33 150 50 2.4 Below pipe [??] [??] [??] [??] [??] [??] Actual data from the Model output validation set Predicted model condition Pipeline (No.) Actual CCTV Model value condition 1 2 8.3 2 2 1 7.8 2 3 1 7.9 2 4 1 8.3 2 [??] [??] [??] [??] 465 3 6.2 3 466 2 6.1 3 [??] [??] [??] [??]

Printer friendly Cite/link Email Feedback | |

Author: | Hawari, Alaa; Alkadour, Firas; Elmasry, Mohamed; Zayed, Tarek |
---|---|

Publication: | Australian Journal of Civil Engineering |

Geographic Code: | 7QATA |

Date: | Apr 1, 2018 |

Words: | 7830 |

Previous Article: | Haul road rolling resistance and pavement condition. |

Next Article: | Laboratory evaluation of the characteristics of continuously reinforced concrete pavement incorporating recycled concrete aggregate. |

Topics: |