Construction Innovation Research Center regarding Consulting Services at its own risk. Construction Innovation expressly disclaims any responsibility or liability to any person in respect of anything done or omitted to be done by any person in reliance on this Report or any information provided. Based on this documentation, a software tool (BridgeDIST) with a "knowledge base" and an inference engine has been developed.
A hard copy costing module has been developed to accompany the software tool that evaluates the cost of recovery methods for a given scenario.
Introduction
Background
For readers unfamiliar with the above mechanism descriptions, QDMR's Bridge Inspection Manual (BIM) may serve as an informative reference. The following part of the report presents a brief summary of the research objectives, methodology and strategies. The details of the two case study bridges selected from the above table are presented in the next section.
For privacy reasons, the names and locations of the case study bridges have been omitted.
Bridge 1
- Bridge details
 - Observed Distress
 - Distress in Pile caps
 - Distress in Piles
 - Hypothesis
 - Tests undertaken
 - Diagnosis
 - Prediction of the dominant mechanism causing distress
 - Pile caps
 - Piles
 - Prediction summary
 - Proposed Remedial actions
 - Option 1: Leave `as-is’
 - Option 2: Cathodic protection & repair of pilecaps; Encasement of piles
 
Tapping with a hammer showed that a large part of the cover concrete had delaminated in these places. It was observed that all piles had vertical cracks on almost every surface of the pile. It was observed that the delaminations observed in the cover concrete were due to expansive corrosion of the reinforcement.
This is further confirmed based on the possibility of double diffusion of chlorides from two adjacent sides of the pile caps. The type of cracks observed underwater in the piles, their appearance and the age of the bridge were indicative of Alkali-Silica Reaction (ASR). The concrete quality was estimated to be approximately 20 MPa based on test results and the age of the structure.
It is clear that the chloride content is approximately twice the theoretical threshold required for chloride-induced corrosion to occur. This was expected given the aggressive tidal environment of the structure, its age and the low grade of concrete used in the pile caps. The extent of cracking and delamination observed in the pile cap inspection video indicates that the mechanism is widespread and affects all pile caps.
The effects of ASR in the pile caps are believed to be relatively minor, based on the findings of the petrographic report. Eventually, significant loss of diameter of the steel and failure of the pile caps is likely to occur. The investigation phase concluded that the cracks in the post caps were due to extensive corrosion of the reinforcing steel, accelerated by the tidal environment.
With the mechanism causing steel corrosion controlled by the CP system, the new repairs would ensure a sound condition of the pile caps.
Expert systems for general concrete structures
Experts systems on transition probability models
Expert systems on prioritising sewer inspections
- List of Elements and components
 - Types of Construction
 - Types of Material
 - Type of aggressive environment
 - List of common defects
 - Cracking
 - Crack width
 - Crack pattern
 - Crack location
 - Crack appearance
 - Crack depth
 - Crack length
 - Crack growth
 - Crack occurrence
 - Environmental data
 - Design data
 - Concrete Cover
 - Concrete Grade
 - Construction data
 - coarse aggregate
 - Fine aggregate (Sand)
 - Cement
 - Exposure classification
 - Process of Chloride ingress
 
Further analysis of the scientific evidence to relate these symptoms to the corresponding mechanisms was undertaken. For example, prestressed elements affected by ASR exhibited cracks along the line of least resistance of the member. This is due to the fact that the expansion due to ASR can initiate the formation of gels leading to the cracking of the aggregates or the cement paste, in which case the elements that are already prestressed, only along the line of least resistance will crack.
The reason for this cracking in the card could be that the ends are unrestrained and thus the expansion and cracking would try to follow the reinforcement pattern, while due to the restraint a typical “card cracking” pattern could be observed. In chloride-induced corrosion, cracking occurred only within the concrete cover and was clearly accompanied by staining, delamination, and peeling. This is due to the fact that the penetration and accumulation of the chlorides would depassivate the protective film of the reinforcement in the concrete.
Based on the detailed analysis a complete mind map including all dominant mechanisms was developed for the software tool. 2 Inadequate compliance with codes of practice (eg AS 5100) 3 Poor classification due to incorrect water cement ratio. 4 Poor grade due to unsuitable aggregates 5 Poor grade due to non-uniform compaction 6 Poor grade due to non-uniform reinforcement 7 Poor grade due to other reasons 8 Adequate or good grade.
The above tables clearly represent the number of variables used in the development of the software diagnostic tool. The approach used in the development of the software inference engine is presented in the next section.
Modelling approach
Reasoning engine
A linguistic variable must have valid syntax and semantics, which can then be specified by fuzzy sets or rules. A syntactic rule defines the well-formed expressions in T(L), where the term T(L) is a set of linguistic variables and L is the set of values it can take. Initially, the software checks for the presence of the variable “Element” (i.e. a stack, stack cap, columns, etc.).
It is mandatory to select an element; otherwise the software will not continue to the next screen. Once the user selects an element and enters the available information, all rule bases associated with that element are taken up for assessment. Then each of the fuzzy vectors is compared with the subsequent vector so that if F(i) < F(i+1) then the vector F(i+1) is selected.
Note that a weight vector is not required here, as the comparison is based on user input as a reference point. However, it will be useful to assign importance vectors to each of the variables 'Vi' in the developed rule base, currently under consideration by the authors. If the situation arises where the difference between two vectors is zero (which is unusual), they are both included for further evaluation with the remaining vectors.
In these situations, the user is offered three matching solutions, with a certain degree of confidence attached to each assessment. In the event that two or three mechanisms with close confidence limits appear to match the input, the matching mechanisms, based on the percentage of their input, are shown to the user with the advice that further accurate information is required or the software needs to be rerun performed to complete the input. Evaluate the dominant emergency mechanism.
Software development
Trial run of the software tool
Note that the order of the variables shown here corresponds to that developed in the software and the software would have compared all the rule bases that exist in the database. V2 Construction Prestressed Very High V3 Material Sit-in-situ Very high V4 Environment Saltwater Very high. For the element that is Pile, 3 different rule bases relevant to ASR, Chloride Induced Corrosion and Construction are selected from the database for comparison.
V71 Crack pattern (C) Vertical, flowing above and below water V72 Crack location (C) Sections where concrete cover is less (eg mid-face of polygonal columns) V73 Crack width (NC) Line of hair to medium. V2 Construction (C) Reinforced concrete V3 Material (NC) Cast in place. position) (C) Submerged / tidal area V71 Crack pattern (C) Initially horizontal V72 Crack location (C) Longitudinal V73 Crack width (NC) Hairline to medium. V8 Concrete Grade (C) Mostly Unsuitable Concrete Cover V9 (C) Very Unsuitable V10 Coarse Aggregate (NC) Medium to Good Quality V11 Fine Aggregate (NC) Medium to Good Quality V12 Cement Type (NC) Medium to good quality V13.
P2 Construction (C) Hollow Spinning P3 Materials (NC) In-situ P4 Environment (Q) Domestic P5 Climate (Q) Temperate P6. V) Away from water influence V71 Crack pattern (C) Horizontal / circumferential V72 Crack location (C) Along the length. Q8 Concrete Grade (Q) Sufficient / Insufficient Q9 Concrete Cover (Q) Sufficient / Insufficient Q10 Coarse Aggregate (NC) Medium to Good Quality Q11 Fine Aggregate (NC) Medium to Good Quality Q12 Type of Cement (NC) Medium to Good Quality Q13 . Comparing the user input with the 3 rule bases in accordance with the membership equations 5 to 10 results in the following values:
Thus, the fuzzy vector corresponding to rule base 1 is chosen as the best possible matching mechanism. After choosing fuzzy vector F1 as the best matching mechanism, the degree of confidence is calculated taking into account the degree of confirmation of the user's input as follows:.
Software details
Screen dumps of the software
The following screen is the interface where the new rules are defined (part of the expert user interface; not part of the general user interface). For example, crack details like crack pattern, crack width are entered here. The tool presents a degree of confidence for predictions based on the amount and quality of information.
An example of the typical rule-based matrix developed in this research is presented in the table below (in three parts and to be read in context). This report has presented extensive documentation of the development of the knowledge base and the inference engine that includes a rule-based matrix approach to the development of a diagnostic tool "BridgeDIST". The developed software is open, and further development of the rule base can be carried out or input by expert users.
This report is part of the CRC CI Research funded research project entitled "Sustainable Infrastructure for Aggressive Environments". Risk Management to Ensure Long Term Performance in Civil Infrastructure” 21st Biennial Conference of the concrete institute of Australia, Brisbane, Australia, pp. Condition evaluation of concrete bridges relative to armament corrosion, Volume-1, State of the art of existing methods”, Report Number SHRP-S/FR-92-103, National Research Council, Washington DC, 1996.
A framework for predicting the residual load carrying capacity of concrete structures exposed to aggressive environments”, Proceedings of the 3rd International Conference on Construction Materials: Performance, Innovations and structural implications, Aug, 2005. An expert system for evaluating the emergency mechanisms in bridges exposed for aggressive environments", Journal of Concrete Institute of Australia, (submitted for publication).