STRENGTH EVALUATION OF UNBONDED POST-TENSIONED STRUCTURES WITH SEVERE TENDON BREAKAGE Norman R. Webster, M. Eng., P. Eng. Professor David M. Rogowsky, Ph.D., P. Eng. Read Jones Christoffersen Ltd. Department of Civil & Environmental Engineering #200 Kensington Road, N.W. University of Alberta Calgary, Alberta, CANADA 220 Civil Engineering Building T2N 3P9 Edmonton, Alberta, CANADA T6G 2G7 KEYWORDS: Corrosion, Unbonded, Post-Tensioning, Strength, Evaluation, Reliability, Partial Prestressing, Prestressed. ABSTRACT: Many unbonded post-tensioned structures in Canada, and also in some areas in the U.S.A., constructed prior to the late 1980’s have exhibited extensive tendon breakage caused by water ingress and corrosion. Evaluation of such structures is needed to determine strength and safety at the time of investigation, and to determine what degree of breakage can be tolerated before shoring or repair is required. The problem presents significant challenges to the structural engineer for the following reasons. The problem is relatively new. Methods of analysis used for new design are not always applicable, especially in cases of severe breakage. Codes and consensus guidelines for such analyses have not been extensively developed. Structures can behave differently at low levels of prestress (i.e. after severe breakage) than new structures designed according to modern codes. Field testing for breakage is costly and disruptive and is therefore often limited in extent. There are many variables within a structure and many variables between structures (loading, material properties, amount and reliability of test data, amount of breakage, variability of breakage within a structure, original construction tolerances, original level of effective prestress.) The ductility of structures with minimal reinforcing and low prestress has not been extensively studied (and hence the likelihood of pre-collapse deflection and warning is not always well defined). Yet within these limitations, the structural engineer must form an opinion on the strength and safety of structures with severe breakage. The paper will present a strength evaluation method that considers the above mentioned variables. Load testing of samples with very low prestess will also be presented. INTRODUCTION Problem definition Many unbonded post-tensioned structures in Canada, and also in some areas of the U.S.A., constructed particularly prior to the late 1980’s, have exhibited extensive tendon breakage caused by water ingress. Most, but not all, of the instances of severe deterioration in typical post-tensioned structures in Canada have been found in systems commonly referred to as the “Push-Through” type (Figure 1). In such systems, the tendon was assembled by inserting a greased strand into a loose-fitting plastic sheath, leaving an air space between the sheath and the strand. The intent of such systems was that the plastic sheathing and the grease coating were to prevent bond of the concrete to the tendon, and to reduce friction to make post-tensioning possible. Some amount of corrosion protection was provided by the grease, plastic, grout plugs and concrete cover. However, the systems failed to provide corrosion protection as a result of several characteristics. Water could enter the sheathing through the end of the anchor itself along the axis of the strand, or into the end of the plastic sheathing adjacent to the anchor, or through perforations in the sheathing caused by impact or Figure 1 – Cross-section of Typical Unbonded Tendon with Loose Sheathing (“Pushed-through” System). abrasion during fabrication, transportation or construction. Water also penetrated after construction through the concrete (and then through perforations in the sheathing), and through grout plugs. De-icer salts (chlorides), air borne seawater, and fertilizer from landscaping have been found to aggravate the corrosion situation. However, in many cases, pure water has been found to be sufficient to cause corrosion and eventual breakage of tendons. Some instances of deterioration have been found with extruded tendons. However they were less frequently used than the “push-through” systems in Canada during the time period in question and they are somewhat less vulnerable to moisture ingress than the “push-through” systems. The grease coating on the strands was sometimes sparsely applied. Regardless of this, if water gains access the grease may deteriorate leading to corrosion of the strand. Failures observed have typically not been the result of general loss of cross section. Rather, breakage of individual wires, and eventually all 7 wires of the strand, has often resulted from formation of cracks through the wires (Fig. 2), at locations where pitting has been observed in the range of 0 to 5% of the wire cross section (Rogowsky & Robson, 1998). Figure 2 – Typical Fracture Appearance Individual buildings exhibit a broad variation in durability. Some examples are: • Many office buildings, apartments, and parking structures over 15 years old with no breakage. • An office building in Calgary, interior environment (post-tensioning anchors protected from the weather after completion of construction) exhibiting moisture in 24 % of sample tendons exposed at inspection ports (i.e. the actual percent of tendons contaminated by water is probably very high), and 1 % broken or unstressed, at age 12 years. • A parking structure in Calgary, Alberta, Canada, (cold climate) exposed to de-icer salts, exhibiting 3.5 % breakage at age 20 years • An apartment building in Calgary with post-tensioning anchors at or below grade, exposed to water and fertilizers from landscaping, exhibiting 47 % of the tendons in the ground floor slab broken at age 8 years. • An apartment building near Vancouver, Canada, (wet, temperate climate) with post tensioning anchors in exterior bearing walls exposed to rain, exhibiting 5% to 58% breakage in various floor and roof slabs at age 13 years. Because there is usually no external evidence of the corrosion hidden within the structure, one usually needs to actually conduct a physical investigation in order to determine the condition of a structure’s post-tensioning system. In North America, the largest number of investigations has been conducted in Canada and most of the known problems have been detected in Canada. In the USA, problem structures have been identified as a result of investigation deterioration due to de-icer salts in parking structures, or due to corrosion in buildings in a warm seaside environment. In many of these situations, corrosion of non-prestressed reinforcing bars and post tensioning anchors provided visual evidence of problems. However, based on the Canadian experience, one cannot rely on external appearances, and the extent of deterioration outside of the above mentioned environments may be underestimated. In recent years improvements have been made to proprietary post-tensioning systems and to construction techniques. Well-constructed post-tensioning systems utilizing waterproof anchor and sheathing assemblies and fully filled with corrosion inhibiting grease should not experience the problems mentioned above. Breakage of tendons has several possible consequences: • Loss of strength, leading to safety concerns or limitations on load carrying capacity, • Need for engineering evaluation of current condition and long-term case management, even if current condition is satisfactory. • Loss of real estate value may also occur due to repair costs and disruption, whether repairs are needed immediately or whether there is a risk that repairs are needed in the future, and Hazard from eruption of tendons from the top or bottom surface of the structure or from the anchorages (The likelihood of a tendon erupting upon breakage, and the hazard presented in the event of eruption, are both factors that vary widely from one structure to another. This topic is not addressed further in this paper). Evaluating the strength of existing structures presents significant challenges for several reasons. • The problem is relatively new. • Methods of analysis used for design of new structures are not always applicable to existing structures, especially in cases of severe breakage. • Codes and consensus guidelines, in particular for severely deteriorated post-tensioned structures, have not been extensively developed. • There are many variables within a structure and many variables between structures (loading, material properties, amount and reliability of test data, amount of breakage, variability of breakage within a structure, original construction tolerances.) • Field testing for breakage is costly and disruptive and is therefore often limited in extent. • Structures may behave differently at low levels of prestress (i.e. after severe breakage) as compared to new structures designed according to modern codes. • The ductility of structures with minimal reinforcing and low prestress has not been extensively studied (and hence the likelihood of pre-collapse deflection and warning are not always well defined). Yet considering the above challenges, the structural engineer must form an opinion of the strength and safety of any structure with severe breakage. Code Applicability Design codes are intended to provide an appropriate level of reliability for structures that have yet to be constructed. These codes are not necessarily applicable to the strength evaluation of an existing structure for several reasons: • In new construction, incremental increases in safety can be achieved at low cost. Therefore the code load factors are slightly more conservative than warranted by risk analysis, • Minimum reinforcing or minimum levels of prestress may not meet current codes, however the strength may nevertheless be acceptable, • Confirming as-built dimensions and dead loads can remove some of the uncertainty, and hence justify reducing load factors, • Confirming in-situ material properties can remove some of the uncertainty, and hence justify increasing capacity reduction factors or material strengths, • The actual structure may have greater (or lesser) capability to redistribute moments than implied by design codes, and • Long term good performance in actual service conditions can in some cases be justification of acceptability of a structure’s capacity. (However, in cases where the structure is actively deteriorating, past performance is not a reliable indication of future performance). Further information on strength evaluation of existing structures is provided in Commentary K to the National Building Code of Canada 1995 and Clause 12 of CSA Standard S6-1988 . Scope of this Paper Evaluation of deteriorated structures is needed to determine strength and safety at the time of investigation, and to determine what degree of breakage can be tolerated before shoring or repair is needed. This paper will present a strength evaluation method for determining the safe load carrying capacity of severely deteriorated unbonded post tensioned structures for situations where design codes and formulas used for new structures cannot be applied. Considerations for load testing of structures with very low prestress will also be presented. ASSESSMENT OF STRUCTURE CONDITION Before the strength of a structure can be evaluated, an investigation must first be done. Reasons for Investigation Because there is usually no external evidence of deterioration, some other reason us usually necessary to initiate an investigation. Typical examples are: 1. A knowledgeable owner may be aware of the risks of deterioration and commission an investigation without seeing external signs of distress. 2. External evidence, e.g. eruption of a tendon, or corrosion and delamination of concrete due to corrosion of anchors or non-prestressed reinforcement. 3. A knowledgeable purchaser or financier would request an investigation before making financial commitments in order to understand the risk of short term or long term repair costs. 4. Investigations can be costly and disruptive. Therefore initial investigations done for the above- mentioned reasons are usually preliminary in nature. If the preliminary investigation indicates a concern, further investigation would be done to define the extent of breakage or to further define the risk of future deterioration. 5. If there is a significant risk of current or future breakage, then the structural engineer would need to determine: • Is the structure presently safe enough that no load restrictions or shoring are required? and if so, • What amount of breakage can be tolerated before shoring, load restrictions, and/or repair are required? Inspection Techniques The techniques used in evaluating the condition of unbonded post-tensioned structures are discussed by Harder and Rogowsky, 1999 and include the following: 1. Visual inspection to make a general assessment of risk that deterioration has occurred or will occur. 2. Inspection of short sections of a sample of strands, at inspection ports chipped into the underside of the structure. (A “penetration test” performed by driving a screwdriver blade between the wires can indicate a lack of tension in the individual wires of the strands. Strand deflection devices commercially available can also be used to assess total tension in strands) 3. Extraction of full length sample tendons for visual and laboratory inspection. 4. Laboratory testing of sample materials from extracted tendons. 5. Monitoring – either periodic monitoring of tendon tension at sample inspection ports (see item 2 above), or acoustic monitoring (Paulson, 1996) which can detect breakage of individual wires. 6. Measurement and testing of as-built dimensions, loads, and material properties. Obviously a critical aspect of strength evaluation is the determination of the number of broken tendons. This is discussed later in this paper. GENERAL CONSIDERATIONS FOR STRENGTH EVALUATION OF STRUCTURES Before discussing issues that are peculiar to unbonded post-tensioning, it is worth discussing evaluation issues that are generally applicable to all structures. In limit states design vocabulary, the capacity (resistance) must exceed the demand (load). Capacity and demand are discussed separately. Confirm Demand Side of Ultimate Limit State Equation In typical concrete structures, the dead load exceeds the live load. Confirm the dead load by measuring concrete density and dimensions. Confirm partition loadings. The partition allowance used in the original design may be in error. Confirm the live loads due to use and occupancy. The occupancies may have changed. Furthermore, in structures such as parking garages for passenger cars, an evaluation of the maximum plausible live load may be warranted. For example with current passenger car weights and dimensions, one cannot produce a load greater than about 1.5 kPa. The usually specified live load for such a parking garage in Canada, and the US is typically 2.4 kPa. The structural analysis that converts the loads to load effects such as bending moments and shears needs to be considered carefully. In new construction, one may base the design on a simplified conservative structural analysis without significant financial penalty. When evaluating a structure, the cost of unnecessary conservatism can be substantial if it triggers an unnecessary repair. At the ultimate limit state, cracking is substantial and significant moment redistribution may take place. Such a state is better represented by a plastic collapse analysis than an elastic analysis. Figure 3 illustrates example mechanisms for a given deflection δ, the increase in tendon length ∆L can be found from geometry. Dividing ∆L by the free length of the unbonded tendon, from anchor to anchor gives the increase in tendon strain which in turn is used to calculate the change in tendon stress, tendon force and internal moments at each hinge location. One can then develop a load deflection curve as shown in Figure 4. Depending on the specifics of the problem,failures may occur with little or no warning (point b, or between points e and f). Many existing structures will reach point g and have deflections of at least: δ= L 100 – L dp where L is the span and dp is the effective depth of the prestressed reinforcement (Rogowsky and Daher 1999). It should be noted that, if the structure reaches point g, considerable moment redistribution will have occurred. In such a case, the moment demand at any particular cross-section obtained from an elastic analysis is not particularly relevant for checking of the flexural ultimate limit state. In general, it is legitimate to adjust load factors based on the reduction in uncertainty of loading, and consequences of failure. This process was recently formalized in Structural Commentary K to the National Building Code of Canada, 1995. The risk category, system behavior, and measurement of dead load are given numerical values which are summed to establish the reliability level and corresponding dead and live load factors. Tables 1, 2 and 3 are simplifications of those found in Commentary K. For example, if one is evaluating a parking garage where: fewer than 100 people are usually present at any given time; the governing failure mode will be a ductile flexural failure of the slab under a single vehicle such that it is unlikely to cause serious personal injury; and the dead load has been measured, the reliability level = 1 + 1 + 0 = 2. The dead and live load factors for structural evaluation purposes are 1.11 and 1.2 respectively. For comparison purposes the dead and live load factors in new construction in Canada are 1.25 and 1.5. Figure 3 (a) – Simple Span With Mid-Span Hinge Figure 3 (b) – Continuous Span with Mid-Span Hinge For other hinge locations and conditions, calculate ∆L from the geometry of the mechanism. FIGURE 3 – Mechanism Method for Predicting Change in Tendon Force Figure 4 – Idealized Response for Lightly Prestressed Unbonded Concrete Flexural Members TABLE 1 RISK CATEGORY CATEGORY DESCRIPTIONS High Schools and other occupancies where more than 100 people are exposed to risk of failure at any given point in time, buildings of major importance. Medium Other occupancies where 5 to 100 people are exposed to risk associated with failure. Low Other occupancies where fewer than 5 people are exposed to risk at any given point in time. TABLE 2 INDICES FOR THE CALCULATION OF RELIABILITY LEVEL ITEM INDEX Risk Category High 2 Medium 1 Low 0 System Behaviour Failure leads to collapse, likely to 2 impact people Failure unlikely to lead to collapse or 1 unlikely to impact people Failure local only, very unlikely to 0 impact people Dead Load Not measured 1 Measured 0 TABLE 3 LOAD FACTORS FOR STRUCTURAL EVALUATION LOAD FACTORS RELIABILITY DEAD (1) LIVE LEVEL 5 1.25 (0.85) 1.50 4 1.20 (0.88) 1.40 3 1.15 (0.91) 1.30 2 1.11 (0.93) 1.20 1 or 0 1.08 (0.95) 1.10 (1) Value in brackets applies when dead load resists failure. Confirm Capacity Side of Ultimate Limit State Equation The capacity, or load resistance, depends on the in-place material properties, concrete dimensions, as well as size and location of reinforcement. While information from drawings is useful for preliminary calculations, it should be confirmed by field measurements and testing of material samples. It is not uncommon to find that the drawings do not represent what was actually built. If one has a large number of material tests, the characteristic material strengths can be established reliably from the statistical distribution. If one has a limited number of material tests to establish the current strength of the concrete and reinforcement, it is advantageous to use Baysian updating to establish the appropriate values to use with conventional material resistance factors. On the basis of experience and judgment, the evaluator selects what the material strengths are likely to be. A limited number of tests are conducted, and based on the results, the estimated strengths are modified using Bayes’ theorem. The process incorporates the evaluator’s knowledge, judgement and experience as well as the additional information provided by the limited number of tests. For further details see Supplement No. 1 of CSA Standard S6-1988 which includes examples of a priori strength distributions and how to use them to determine the appropriate strength value for predicting factored member resistance. When assessing the capacity of a statically indeterminate member the capacity is not limited by the formation of a single plastic hinge. One needs to consider the possible collapse mechanisms and needs to ensure that the first hinges to form have enough rotation capacity for the last hinge to develop. One may do this explicitly, or one may conservatively use the concept of moment redistribution to modify results of elastic analysis. The discussion of Figures 3 and 4 dealt with this. Special Considerations for Strength Evaluation of Unbonded Post-Tensioned Structures The special considerations for unbonded post-tensioned structures deal primarily with the capacity side of the ultimate limit state equation. The current tension in the tendon is the most important parameter in the flexural strength of a member with unbonded tendons. This is usually extremely difficult to establish without a lift-off test, or a cut-wire test. The next challenge is identifying areas were there is a concentration of tendon breakage. The screwdriver penetration test will usually identify tendons, which are broken, or have serious tension deficiencies. The analysis should start with bays which have the most damage, and progress toward areas with less damage. In this way, one can delineate areas, which require remedial action from those areas, which are adequate. When tendons with one or two out of the seven broken wires are encountered, one must decide if the remaining unbroken wires should be included in the capacity analysis. If the agents which caused the initial wire breaks persist, presumably all of the wires will eventually break. Commonly, if a broken wire is found, the entire seven-wire strand is ignored in the strength calculations. Appropriate sample size is important. If the design is very robust, the breakage rate in the sample is low and the structure can tolerate a high percentage of the tendons broken, then the number of tendons that are inspected and tested for tension with the penetration test need not be very large to determine if safety is acceptable. On the other hand, if the design can tolerate only a small percentage of broken tendons, a large portion of the tendons need to be tested for tension to ensure that there are no areas with unacceptably high breakage rates. The breakage rate found in the sample will also significantly effect the sample size that is needed to confirm safety. See Pandey and Nessim, 1996 for further information on sampling. If the structure has a low tolerance to breakage, if the average breakage rate is high, and if corrosion is systematic, the sampling rate may need to approach 100%. The anecdotal evidence suggests that if the corrosion is due to systematic deficiencies in design, materials or workmanship, the breakage rates will be high and concentrated in areas with the systematic deficiencies. On the other hand, if the deficiencies are due to random causes such as accidental tears in the plastic sheath, the breakage rates will be low with breaks randomly distributed throughout the structure. The force from the post-tensioning anchors tends to spread out as one move away from the anchor. That is, the in-plane compression stresses in the concrete due to the prestressing spread more or less uniformly across the slab, even if there are minor areas with broken tendons. The area with the broken tendons will have lost the load balancing effects from the broken tendons. Locally, this can be very significant. The loss of in-plane stress from the broken tendons will be spread out over the entire slab width at an angle of perhaps 2 to 1. Thus all areas see some loss of prestress. This is important when one determines the collapse load for a bay or panel of floor slab. One must consider the possibility of shear failure as well as flexural failure. The procedure illustrated in Rogowsky and Daher (1997) can be used to assess the capacity of a flexural failure mechanism and determine if it will be ductile or brittle. Miltonberg (1998) extended the work to include the effects of in-plane restraint. This latter effect increases the strength of the system through arching action, but reduces the deflection before failure. Arching action is particularly important where local areas experience a concentration of broken tenders. Repair Strategies and Techniques In the event that repair is required, several options exist: • Removal and replacement of all strands • Monitoring (manually or acoustically) and replacing strands only sufficient to meet near term structural requirements • Structural steel sub-frame • External reinforcing: bonding or bolted plates or composite materials • External post tensioning • Demolition and reconstruction • Combination of the above. Load Testing Load testing is one possible strategy that may be employed. For buildings with corroding unbonded post-tensioning it is not usually an attractive strategy. While it can confirm the structures ability to carry a particular load at the time of the test, it does not shed much light on the strength with future corrosion and tendon breakage. If the structure does not develop flexural cracks during the test, one could be rather close to failure and not know it. The load test would stop before point b on Figure 4, and no information on the structure’s strength and durability would be obtained. If this is the situation, one should consider increasing the proof load as one does for brittle materials like glass. The Need for Judgement Each structure’s situation is unique: its structural configuration, current condition, prognosis for future breakage, owner’s requirements, localized vs. random/distributed breakage, capability for plastic redistribution of moments, etc. Therefore, the matter of evaluating a severely deteriorated structure is one that lends itself to a recipe approach. It is a matter that demands the application of wise judgement and extensive experience and insight into the behaviour of concrete structures. REFERENCES: Canadian Standards Association (1988), CSA Standard S-6 – 1988, Design of Highway Bridges. Harder, J. & Rogowsky, D., (1999), Inspection and Monitoring of Buildings with Unbonded Prestressing, Presented at Structural Faults + Repair – 99, July 13-15, 1999, Commonwealth Institute, London, UK. National Research Council of Canada (1996), Ottawa – User’s Guide – National Building Code 1995, Structural Commentaries (Part 4), First Edition 1996, ISBN 0-660-16275-X. Pandey, M.D. and Nessim, M.A. (1996), Reliability Based Inspection of Post-Tensioned Concrete Slabs, Canadian Journal of Civil Engineering, Vol. 23, 1996, p.p. 242 – 249. Paulsen, P.O. (1996), Continuous Acoustic Non-destructive Evaluation of Unbonded Post-Tensioned Strands, Third Conference on Non-destructive Evaluation of Civil Structures and Materials, September 1996. Rogowsky, D. & Daher, Y (1997), Flexural Behaviour of Monstrand Buildings with Corroded Tendons”, Symposium on Advances in Design, Construction, Evaluation and Monitoring of Structures, June 12 – 13, 1997, Calgary, Alberta Rogowsky, D & Robson, N. (1998), Influence of Corrosion and Cracking on Tendon Strength and Ductility, Workshop on Environment-Assisted Cracking of Unbonded Post-Tensioned Tendons, June 18-19, 1998, Calgary, Alberta. STRUCTURAL FAULTS + REPAIRS – 99 AUTHORS BIOGRAPHICAL DETAILS TITLE OF PAPER: STRENGTH EVALUATION OF UNBONDED POST-TENSIONED STRUCTURES WITH SEVERE TENDON BREAKAGE NAME OF AUTHOR: Norman R. Webster QUALIFICATIONS: B. A. Sc., University of British Columbia (Honours) 1972 M. Eng., University of British Columbia 1975 P. Eng., (Professional Engineer), Alberta, Canada CURRENT EMPLOYER/ORGANIZATION & POSITION: President, Read Jones Christoffersen Ltd. BRIEF CAREER SUMMARY: Read Jones Christoffersen Ltd., 1973 – Present Involved in evaluation and restoration of post-tensioned structures since 1985. SPECIAL AWARDS/PRIZES, DECORATIONS, ETC.: International Parking Institute – 3 Awards of Excellence and 3 Awards of Merit for design of various parking structures (1985 – 1998). MEMBERSHIP OF TECHNICAL COMMITTEES: Canadian Standards Association, CSA Committee S-413, Parking Structures (Vice-Chair) ACI International Committees 362 (Parking Structures), 423 (Prestressed Concrete), and 437 (Strength Evaluation) PUBLICATIONS: Webster, N.R., (1989), Unbonded Monostrand Post-tensioned Construction – A Question of Durability, Presentation to Canadian Society of Civil Engineers, Vancouver, B.C., June 21, 1989. Webster, N.R. (1991), Evaluation of Unbonded Post-tensioned Structures, Presentation to 2nd Canadian Symposium of Cement and Concrete at the University of British Columbia, Vancouver, B.C., July 25 – 26, 1991. Webster, N.R. (1993), Corrosion in Unbonded Post-tensioned Structures: History, Safety, Causes, Evaluation, Repair, Presented to the City of Calgary Seminar on Safety of Unbonded Post-tensioned Structures, November 25, 1993. Webster, N.R. (1994), Evaluation Techniques and Building Performance Related to Corrosion in Unbonded Post-tensioned Structures, Presented to ACI-Alberta Chapter Fall Convention in Calgary, November 15, 1994 and in Edmonton, November 16, 1994.