Correct defect identification and damage assessment in general and of corrosion defects in particular as basis for efficient sewer rehabilitation planning

1 Introduction
Operation, maintenance and rehabilitation of sewer networks are mainly driven by the results of CCTV inspections. A standard classification scheme has been issued with the European EN 13508 code. The inspection results are of course suitable for prioritising the most urgent measures within the network, because of leaks exfiltrating fowl water to the aquifer or collapses causing instability, which are detected in a pipe. However, correct identification of certain damage types which change over time such as corrosion, does gives the opportunity for a pro-active rehabilitation of the network and eases the identification of the causes for such deterioration. Therefore the three steps identification, assessment of current state and prediction of aging behaviour are closely mended together as basis for an efficient rehabilitation planning as it is shown in figure 1.

2 Defect identification
The first step but probably the hardest one is the correct identification of the damages in general and of corrosion damages in particular. Whereas defects like fissures or breaks are more easily to determine, corrosion defects are likely to look like abrasion defects and may hide, even after cleaning, beneath incrustation and sedimentation. One indicator to separate abrasion from corrosion defects is the circumferential location of the defect as abrasion is most likely to happen on the invert of the sewer pipe. As shown in figure 2, there are several possible positions for pipe surface defects, but only for type c) a closer look needs to differentiate between the two possibilities for the others it are most likely corrosion defects.

Type a) and b) are on the crown and because there is no defect at the invert it must be corrosion, for type d) the decision is not that obvious but if the defect stretches from the invert closed to pipe crown it is most likely to be corrosion as well.

The figure 3 illustrates a brick sewer with defective grooves and because the circumferential location of the defect is at the pipe crown, the cause is corrosion. Another indicator for the decision whether the surface damage is caused by abrasion or corrosion is the appearance of the defective surface itself. Especially for concrete sewers this helps to answer the question easily as the mineral aggregates within the concrete are often far less affected by corrosion than the cement. The larger aggregates of the defect indications shown at figure 4 and figure 5 seem to be almost intact whereas the cement between them has been disappeared together with the smaller aggregates - a quite good indication for corrosion. Abrasion would have affected the larger aggregates as well.

For other materials like the damaged reinforced plastic pipe shown at figure 6 the decision is not that easy. The knowledge on the type of sewage is another indicator for the correct identification of the defect. This knowledge is not only helpful for the correct identification of surface damages but is almost mandatory for the rehabilitation decisions at the end of the rehabilitation planning process.

3 Damage assessment
In principle, the identified defects are assigned to damage classes according to their observed structural and functional integrity. Common grading schemes distinguish between 4 and 6 damage classes, which are defined by threshold values for particular types of defects.

These thresholds either base on structural defects or functional deficiencies and their environmental impacts, or both. The schemes vary between simple assignment of grades to inspection codes and more complex scoring models.

Due to its binary logic, the assignment of damage classes according to strict thresholds leads to ascend of a defect into the next higher class and thus to a precipitous condition degradation, if the defined threshold value is exceeded, although the absolute change is rather small. Apart from this the thresholds often do not include the various ancillary conditions. As a consequence the assigned damage classes do often not reflect the true impact of a defect on stability, operation and environment.

The framework STATUS developed by Stein & Partner does solve these issues by introducing fuzzy logic to the sewer assessment and by the modern and sophisticated damage assessment models.

3.1 Fuzzy logic within STATUS
By applying fuzzy logic instead of the binary logic, strict jumps into the next classes are smoothened by continuous damage grade transitions. These damage grade transitions are characterized by the degree of membership of the appropriate element to the neighbouring classes.

Thus, the membership degree of an element to a class can be described in form of continuous membership functions infinitely from 0 (no affiliation) to 1 (full affiliation). The fuzzy grading is used within STATUS for all steps of a sewer network assessment forming the mathematical background of the assessment, stretching from damage assessment over pipe evaluation to the forecasting model.

The basic procedure is illustrated in figure 7, demonstrating exemplarily the grading of longitudinal cracks, whereby the grade thresholds from the German standard "Arbeitshilfen Abwasser" act as input variables for the membership functions. A longitudinal crack with the tear width of 2 mm according to this standard as well as for the fuzzy grading belongs to damage grade 3.

However, a longitudinal crack with the tear width 1.5 mm, following the German standard belongs fully to damage grade 2, whereas the fuzzy grading system assigns a portion of 35 % to damage grade 2 and the substantially larger portion of 65 % to damage grade 3, resulting in a final grade of 2.65

Therefore, the available inspection data is considered by these smooth condition transition functions with their numeric value, without giving up the concept of the condition grades as such. In case of lack of exact data, descriptive values such as "high" or "low" can be used instead of numerical ones.

The fuzzy logic system of "STATUS Sewer" fuzzificates the values for the further processing using specific fuzzy membership functions. The fuzzificated values are used within an inference matrix for further processing, the inference matrices can be multidimensional and are build from the fuzzy process rules.

All this leads to a more precise defect evaluation and is the first step towards a more detailed assessment. The concept is adapted to a wide range of defect types, and is suitable for almost any grading system available.

3.2 Sophisticated damage assessment models within STATUS
So far fuzzy logic is just replacing the strict classification of the standard assessment procedures. In the next step of the evaluation system of STATUS additional parameters are taken into account, such as ground water level, load or soil permeability. Thus results in new damage models for the specific defect types such as corrosion. The scheme of such a new assessment model is visualised at figure 8.

The assessment for corrosion for example includes - additional to damage extend - pipe cover, pipe material, pipe type, traffic load, pipe diameter and type of surrounding soil. Based on the probabilistic safety theory numerous multidimensional fuzzy sets were created. In consequence the assessment results of the specific damages are closer to the real damage severity than the rather loose rating of common systems.

At figure 9 the result fields for a corroded concrete pipe of 400 diameter with defined ancillary conditions is shown. It clearly illustrates that the ancillary conditions, neglected in most common assessment models, do have a significant impact on the damage severity.

Taking these additional variables into account therefore draws a much more detailed picture of the damage itself and feeds the subsequent sewer section assessment with more precise data enhancing accuracy

At figure 10 the consequences of the sophisticated assessment system of STATUS are clearly visible, the standard grading system sorts the different defects into five strict classes, which results in a high stocking within the single classes whereas the STATUS approach gives a smooth and more detailed grading of the defects due to their membership to the single classes.

4 Further steps towards efficient rehabilitation planning
4.1 Section Assessment
After the detailed damage assessment STATUS uses a fuzzy logic enhanced section assessment, which is divided into condition assessment and the assessment of the substance. In general the condition of a section indicates the priority of action whereas the substance indicates the type of action required such as repair, relining or replacement, supporting network operators in their rehabilitation decisions.

Both indicators are derived from the single defects located in the related sewer section. This shows clearly, that the quality of this subsequent analytical step is directly dependent on the quality of the defect assessment done before.

4.2 Condition forecasting
The section assessment reveals the present situation of the single sections within the network regarding their structural ability, quality of service and environmental impact. Nevertheless it is just a view on fixed point in time. By projecting the determined present state of the network into the future using strategic fore­casts the network operators get detailed information on the deterioration processes that alter condition and substance of the single sewer sections.

The primary mathematical model behind the forecasts is the model of Markov-chains, which determine the probability of condition changes. Only now the network operators are able to develop efficient long-term rehabilitation plans and determine stable long-ranging strategies for network development, network management, determine future quality of service or coordinate efficiently with other infrastructure network maintenance such as road construction or water network rehabilitation.

5 Conclusion
A sophisticated defect assessment like it is implemented with STATUS is the indispensable basis for an efficient and sustainable management of sewer networks. Especially the assessment of corrosion defects, which impact stability and functionality and affect usually the whole length of a section, needs to be as exact and realistic as possible.

Otherwise risks and costs are increasing and, far more dangerous, become unpredictable. STATUS as a learning and calibratable system gives the opportunity of determination of optimal inspection, cleaning and rehabilitation schedules as well as optimal rehabilitation methods on the basis of an extended sewer section assessment concept and the extensive consideration of boundary conditions.

6 References

Herz, R. and Lipkow, A. (2002): Life cycle assessment of water mains and sewers. Water Science and Technology: Water Supply, Vol. 2 No. 4 pp 51-72. Held on 2nd IWA-World Water Congress Berlin 10/2001. IWA Publishing London 2002.

Herz, R. (1994): Alterung und Erneuerung von Infrastrukturbeständen – ein Kohortenüberlebensmodell. Jahrbuch für Regionalwissenschaft, 14/15, 9-28.

Kleiner, Y (2001): Sceduling inspection and renewal of large infrastructure assets. J Infrastructure systems, 12, 136-143

Kropp, I.; Herz, R. (2005): Schadensprognosemodelle für die Zustandsbewertung von Leitungsnetzen der Wasserversorgung. Wasserwirtschaft, Wassertechnik No. 5 S. 10 ff, Berlin 2005.

Kropp, I.; Lipkow, A. (2002): Erneuerung städtischer Infrastrukturnetze - Vorhersage des Erneuerungsbedarfs und Analyse von Rohrnetzerneuerungsstrategien. in: GeoBit 10/ 2002, 18ff. Heidelberg, 2002.

Le Gat, Y. (2004): Degradation Models for Drain and Sewer Pipelines Parameterised with CCTV Inspection Data. Cemagref Bordeaux, Internal paper, unpublished

Müller, N. (1998): Kosten - Nutzen -Untersuchung vorbeugender Rohrnetzerneuerung, dargestellt am Beispiel des Wasserrohrnetzes der Stadtwerke Chemnitz, Diplomarbeit, Lehrstuhl Stadtbauwesen, TU Dresden.

Stein, R.; Trujillo Álvarez, R.; Lipkow, A. (2004): STATUS Sewer – a holistic model for efficient net management based on section specific aging prognoses. NODIG 2004, Hamburg, Germany

Trujillo Álvarez, R.(1995): Bedarfsprognose und Strategieentwicklung für die Rehabilitation städtischer Wassernetzwerke. Dissertation, Karlsruhe 1995.