Temporary Corrosion Protection and Bond of Prestressing Steel

Temporary Corrosion Protection and Bond of Prestressing Steel

Marti, Peter

The results of laboratory and field tests to investigate the performance of different temporary corrosion protection methods for prestressing steel are presented. It is demonstrated that a particular emulsifiable oil product applied to the prestressing steel showed by far the best corrosion protection behavior. Using this product, a pullout test with a long embedment length was performed on a post-tensioned seven-strand tendon with a plastic duct and compared with a reference test using untreated strands. Compared with the untreated strands, a bond shear stress reduction by a factor of approximately 2.5 was observed. It is shown that, generally, this reduction does not significantly influence the load-deformation response of post-tensioned concrete members and that the emulsifiable oil does not need to be removed before grouting of the tendons.

Keywords: bond; corrosion; duct; durability; grout; post-tensioning; prestressed concrete; prestressing steel; test.

(ProQuest: … denotes formulae omitted.)


One of the most important means to ensure the durability of post-tensioned concrete structures is an effective multilayer protection of the prestressing steel, provided by the surrounding concrete, duct, and grout.1 Because, depending on the construction process, the tendons remain ungrouted for a certain period of time, temporary corrosion protection measures may have to be implemented. Typically, temporary corrosion protection measures are required if tendons manufactured on the construction site remain ungrouted for more than 6 weeks or if they are left ungrouted after being tensioned for more than 2 weeks.2 Whereas there is little guidance on suitable temporary corrosion protection methods, it is generally required that they are not harmful to the prestressing steel, the duct, or the bond between the prestressing steel and the grout.3

Temporary corrosion protection methods include (1) the application of emulsifiable oils to the prestressing steel; (2) the continuous pumping of dry air through the duct to reduce the relative humidity to values below 40 to 50%; and (3) the creation and maintenance of an inert gas atmosphere in the duct using nitrogen.

Emulsifiable oils can easily be applied, are inexpensive, and affect neither the steel nor the alkaline protection provided by the grout.4 Not all emulsifiable oils are suitable for corrosion protection, though, and the protective oil film on the steel can be damaged during transport and tendon installation. Furthermore, emulsifiable oils change the tribology of the steel and thus reduce the bond action between steel and grout; this was confirmed by a number of pullout tests using single strands and wires.5-7

Both the dry air and the inert gas method are rather sophisticated and costly. Installation at the construction site is complex, the system may leak, and there is a danger of condensation. The limited available test data indicate that the methods can be used for temporary corrosion protection up to 1 year.8-10 Recent investigations, however, showed that the same quality of corrosion protection over the same time period can be obtained by applying emulsifiable oil.11

This paper reports on laboratory and field corrosion tests to investigate the performance of different temporary corrosion protection methods. It is demonstrated that a particular emulsifiable oil product, used as a 25% aqueous emulsion, showed a consistently good corrosion protection behavior under all test conditions. Using this product, a bond test on a post-tensioned seven-strand tendon with a plastic duct embedded in a long reinforced concrete prism was performed and compared with a reference test using untreated strands. It is shown that compared with the untreated strands, the average bond shear stresses were reduced by a factor of approximately 2.5. The significance of this finding is discussed.


Four types of laboratory corrosion tests with three emulsifiable oil products in three concentrations resulted in the selection of one superior product with a particular concentration. Field tests during a winter semester on 16 transverse deck post-tensioning tendons of two road bridges demonstrated the practical suitability of the selected product and its superiority compared with other corrosion protection methods for parallel-wire and multi-strand tendons, both with steel and plastic ducts. Two large-scale pullout tests on post-tensioned seven-strand tendons with long embedment lengths provided reliable data to quantify the bond reduction of treated versus untreated strands for the selected product.


Laboratory tests

Laboratory tests under controlled conditions were performed to select an appropriate corrosion-protective agent for the subsequent field and bond tests. Steel plates measuring 100 x 75 x 2 mm (3.94 x 2.95 x 0.08 in.), prestressing wires 7 mm (0.28 in.) in diameter, and prestressing strands 15.7 mm (0.62 in.) in diameter were used as specimens for the laboratory tests7 at the Laboratory for Corrosion and Materials Integrity of the Swiss Federal Laboratories for Materials Testing and Research (Empa). The specimens were dipped into the corrosion-protective agents for 2 minutes, and then they were given time to air dry under standard conditions (23 °C [73.4 °F], 50% relative humidity [RH]) for at least 16 hours. The corrosion-protective agents included Rust-Ban 310, Shellcool M3, and Aseol Milem-23-31, used as 5% and 25% aqueous emulsions and as concentrates (100%).

Four types of tests were carried out, using unprotected specimens as references:

1. Condensing humidity tests with steel plates12: steel plates were subjected to an alternating climate for 3 days (8 hours at 50 °C [122 °F] and 100% RH followed by 16 hours at room temperature and less than 100% RH);

2. Condensing humidity tests with wires and strands13: wires and strands were subjected to an alternating climate for 28 days (8 hours at 40 °C [104 °F] and 100% RH followed by 16 hours at room temperature and less than 100% RH);

3. Semi-immersion tests14 in deionized water: wires and strands were semi-immersed into deionized water with access to air (21 days at 23 °C [73.4 °F], 50% RH); and

4. Semi-immersion tests14 in a saturated calcium-hydroxide solution: wires and strands were semi-immersed into a saturated calcium-hydroxide solution with access to air (21 days at 23 °C [73.4 °F], 50% RH).

After the corrosion tests, the specimens were visually inspected using an optical microscope.

The condensing humidity tests with steel plates showed that Rust-Ban 310 protected the steel from corrosion. Shellcool M3 and Aseol Milem-23-31 provided a moderate corrosion protection. After the tests, many rust spots were visible on the metal surfaces. The unprotected reference specimens were covered with countless rust spots.

The condensing humidity tests with wires and strands demonstrated that the Rust-Ban 310 concentration must be at least 25%. The 5% aqueous emulsion did not protect the steel from corrosion.

This finding was confirmed by the semi-immersion tests in deionized water (refer to Fig. 1(a)). Only Rust-Ban 310 used as a concentrate and as a 25% aqueous emulsion protected the steel from corrosion. The two other agents exhibited a relatively poor or nonexistent corrosion protection, even as concentrates (100%).

The semi-immersion tests in a saturated calcium-hydroxide solution showed no signs of corrosion for Rust-Ban 310 and Aseol Milem-23-31 used as concentrates and little rust formation in the three-phase zone for the 25% aqueous emulsion of Rust-Ban 310. Shellcool M3 and the 25% aqueous emulsion of Aseol Milem-23-31 had no corrosion inhibition effect for the chosen test conditions (refer to Fig. 1(b)).

Because of their high viscosity, concentrated oils are not suitable for practical applications as temporary corrosion-protective agents for post-tensioning tendons. Therefore, based on the results of the laboratory tests that indicated consistently good corrosion protection behavior of Rust-Ban 310 used as a 25% aqueous emulsion, this product was selected for the further bond investigations.

Field tests

Practical exposure tests7 were carried out by the Laboratory for Corrosion and Materials Integrity at Empa from November 2000 to April 2001, using eight transverse deck post-tensioning tendons (each) from the Limmat River bridge at Siggenthal and one of the twin highway bridges at Corcelles in Switzerland (refer to Fig. 2). The tendons were equipped with removable anchor heads at both ends. The tendons were prestressed, left ungrouted for 6 months, and then detensioned, removed, and replaced by new tendons.

For the bridge at Siggenthal, 18 m (59 ft) long parallelwire tendons, each consisting of 22 wires with 7 mm (0.28 in.) diameters were used in combination with three corrosion protection methods, that is:

1. Rust-Ban 310 used as a 25% aqueous emulsion, applied by the supplier of the prestressing wires, and left to dry for 1 day;

2. The use of an adsorption drier for the continuous pumping of dry air through the ducts, while measuring temperature and RH over the whole test period; and

3. Filling the ducts with nitrogen supplied from a steel bottle with compressed nitrogen, pumping for 2 hours every day.

For each of the three methods, one tendon with a corrugated steel duct and one tendon with a plastic duct15 were used. In addition, two reference tendons, one with a steel duct and the other with a plastic duct, contained unprotected wires.

For the bridge at Corcelles, 15 m (49 ft) long tendons, each consisting of four 12.9 mm (0.51 in.) diameter seven-wire strands were used. The strands of four tendons had been treated by the manufacturer immediately after production and before coiling with a 25% aqueous Rust-Ban 310 emulsion. The other four tendons with unprotected strands served as reference. For both the tendons with and without emulsifiable oil treatment, two corrugated steel ducts and two plastic ducts were used.15

The visual inspection of the emulsifiable oil treated tendons after their removal from the bridge at Siggenthal revealed very good corrosion protection, irrespective of the duct material (refer to Fig. 3(a)). Contrary to the laboratory test specimens that were dry within a few days, the wires of the Siggenthal bridge tendons were still coated with an oily layer. The drying behavior is not yet fully understood. The laboratory tests showed that it depends both on the time and the position (laying or hanging) of the steel after application of the emulsifiable oil. On the other hand, experience shows that emulsifiable oil-treated steel coils delivered to construction sites can be wet or dry.

The wires of the tendons that had been protected by the dry air method were dry and showed only a few rust spots. The plastic duct tendon exhibited a somewhat better behavior than the steel duct tendon (refer to Fig. 3(a)).

The wires of the tendons that had been protected by the inert gas method were covered by water drops and showed many rust spots. While the wires in the steel duct behaved approximately as well as their companions that had been protected by the dry air method, the plastic duct tendon’s corrosion behavior was even worse than that of the unprotected reference tendons (refer to Fig. 3(a)). The smell of hydrogen sulfide was noticed as the wires were removed from the plastic duct. Microbiological processes or reducing chemical reactions may have been the reason for the poor corrosion protection behavior of the plastic duct tendon in combination with the inert gas method.

The reference tendons were wet and showed many rust spots over the entire length. Clearly, however, the plastic duct provided a better corrosion protection than the steel duct (refer to Fig. 3(a)).

In summary, the dry air method resulted in a poorer corrosion protection behavior compared with the emulsifiable oil treatment; however, compared with the unprotected reference tendons there was a considerable corrosion protection effect. The inert gas method was approximately as effective as the dry air method for the steel duct tendon and very poor for the plastic duct tendon. This may have partly been caused by the practical difficulties in sealing the ducts and keeping them free of oxygen.

All strands removed from the eight tendons of the bridge at Corcelles were wet. Similar to the bridge at Siggenthal, the emulsifiable oil treatment provided a good corrosion protection, irrespective of the duct material (refer to Fig. 3(b) and 4). Rust spots occurred mainly in the crevices between adjacent wires of the strands. The depth of corrosion attack of the protected strands was less than 50 µm (0.002 in.).


Test setup and measurements

Two identical, prismatic reinforced concrete test specimens were produced and post-tensioned using centrally placed seven-strand tendons with plastic ducts,15 one of them with strands that had been treated by the manufacturer immediately after production with a 25% aqueous Rust-Ban 310 emulsion and the other with untreated strands (refer to Fig. 5(a)). After post-tensioning, the tendons were only grouted over a part of their length, leaving an unbonded length adjacent to the stressing anchorage.

After the grout had hardened, the prestressing force at the stressing end was incrementally increased from its initial value P0 to values of P = P0 + δP. This created additional longitudinal compressive strains δεc in the concrete and the longitudinal nonprestressed reinforcement within the unbonded as well as a part of the bonded length of the tendons. For each δP, the strains were measured on three of the specimen’s surfaces using linearly variable displacement transducers and a mechanical strain gauge (extensometer) with a gauge length of 200 mm (7.87 in.) placed on aluminum targets that had been glued to the concrete surfaces before post-tensioning (refer to Fig. 5(b)). All extensometer readings were taken twice and averaged for each gauge length.

The test method corresponds to a pullout test with a long embedment length. Similar large-scale tests have been used to investigate the differences in the bond behavior of posttensioning tendons with corrugated steel and plastic ducts.16 The two tests described herein belong to a series of largescale tests that were conducted at Empa to investigate the influence of several parameters on the bond behavior of post-tensioning tendons, including the tendon size and shape, the duct material, and the loading sense (loading and unloading).

Test specimens

The test specimens had lengths of 4.87 m (16 ft) and square cross sections with a side length of 280 mm (11 in.) (refer to Fig. 5(a)). They were cast in horizontal position using concrete with a maximum aggregate diameter of 16 mm (0.63 in.). At the time of bond testing, 50 days after casting, the concrete cylinder compressive strength was approximately 49 MPa (7107 psi).

The nonprestressed reinforcement consisted of square stirrups, closed with 135-degree hooks, and four longitudinal 14 mm (0.55 in.) diameter corner bars (refer to Fig. 5(a)).

Instead of uniformly spaced 10 mm (0.39 in.) diameter stirrups, as used for the interior of the specimens, the anchorage zones were equipped with 10 densely spaced 12 mm (0.47 in.) diameter stirrups. Four of these stirrups were rotated by 45 degrees. In this way, the necessary confinement of the anchorage zone, usually provided by a spiral reinforcement, could be obtained from the stirrups.17 The minimum concrete cover of the reinforcement was equal to 15 mm (0.59 in.) and all reinforcement was made from Type B500B steel2 with a characteristic yield strength (5% fractile value) of 500 MPa (72.5 ksi).

Each of the seven 15.7 mm (0.62 in.) diameter, seven-wire strands forming a tendon had a cross-sectional area of 150 mm2 (0.23 in.2) and a characteristic tensile strength (5% fractile value) of 1770 MPa (256.7 ksi), resulting in a characteristic tendon strength of 1859 kN (417.9 kips). The plastic ducts15 had an interior diameter of 58 mm (2.28 in.) and a wall thickness of 2.5 mm (0.10 in.). At each end of the specimens, the strands were anchored using VSL Type E anchorages.15 The stressing anchorage was equipped with a threaded anchor head and a ring nut to keep the forces P = P0 + δP and the associated strains practically constant while taking the strain readings.

The initial prestressing force P0 was equal to 60% of the characteristic tendon strength, that is, 1115 kN (250.7 kips), and it was intended to successively apply force increments δP of 149, 297, and 465 kN (33.4, 66.8, and 104.5 kips), corresponding to 8, 16, and 25% of the characteristic tendon strength, respectively. The force P = P0 + δP was applied by a calibrated hydraulic jack of Type ZPE 1000.15

The tendons with and without emulsifiable oil treatment were grouted 33 and 31 days after the casting of the concrete, respectively, using a conventional grout consisting of Type I portland cement,18 water and a high-range water-reducing admixture. The water-cement ratio was equal to 0.36 and the admixture-cement ratio was 0.01. The specimens were grouted in the vertical position. Due to cement settlement in the grout, the unbonded tendon length increased from 1235 to 1585 mm (48.6 to 62.4 in.) (refer to Fig. 5(a)). At the time of bond testing, the grout had a cube compressive strength of approximately 55 MPa (7977 psi) and a modulus of rupture of 9 MPa (1305 psi).

The temporary corrosion protection was applied by the strand manufacturer by spraying the oil emulsion onto the strands before coiling. Because there was not sufficient time for them to dry, the strands were still wet when the tendons were installed. Concerning bond reduction, compared with untreated strands, this may have represented the most extreme possible conditions.

Test results

Figure 6 shows the development of the average concrete strain increments δεc due to the prestressing force increments δP. It can be seen that due to the highest δP (467 and 463 kN [105.0 and 104.1 kips], respectively), bond lengths of approximately 2.25 and 0.9 m (88.6 and 35.4 in.) were activated, respectively, for the tendons with and without emulsifiable oil treatment. It can also be noticed that the slope of the δεc-x curves within the activated bond length decreases with increasing x and increases with increasing δP for the untreated strands whereas it increases first before decreasing with increasing x for the treated strands.

After having completed the load stage with the highest δP, the tendons were detensioned at the stressing end (P = 0) and the hydraulic jack was removed. This caused a bond reversal within the bonded tendon lengths, resulting in longitudinal cracks in the vicinity of the extensometer targets between x = 0 and x = 0.8 m (31.5 in.) in the specimen containing the untreated strands; no such cracks were observed in the specimen with emulsifiable oil-treated strands, indicating smaller transverse splitting forces due to reduced bond action.


Drying behavior of emulsifiable oils

The bond behavior of prestressing steel depends on its tribology, that is, a better bond behavior can be expected if the emulsifiable oil is dry and waxy instead of being wet and oily. Unfortunately, the condition of an applied emulsifiable oil and its quality can not be controlled yet under practical conditions. The development of a corresponding quality control procedure, including a method to measure the thickness of the oil layer, is highly desirable.

Interpretation of bond test results

Neglecting initial stresses due to creep and shrinkage as well as the stresses due to the self-weight of the test specimen, ignoring the duct and grout stiffness, assuming concrete, reinforcing and prestressing steel to be linearly elastic and simplifying the actual, three-dimensional problem by only considering longitudinal stresses and strains, the bond tests can easily be analyzed.

From Fig. 7(a), with the initial prestressing force P0 = Apσp0, one obtains the concrete stress σc0 = -P0/[Ac – Ad + (ns – 1)As], where Ac, Ad, and As denote the cross-sectional areas of the gross cross section, the duct and the nonprestressed longitudinal reinforcement, respectively, and ns = Es/Ec is the modular ratio of the reinforcing steel and the concrete. With P0 = 1115 kN (250.7 kips), Ac = 2802 = 78,400 mm2 (121.5 in.2), Ad = 632 × π/4 = 3117 mm2 (4.8 in.2), As = 4 × 142 × π/4 = 616 mm2 (0.95 in.2), Es = 205 GPa (29,133 ksi), and Ec = = 36.6 GPa (5308 ksi),2 one obtains ns = 5.6, σc0 = -14.3 MPa (2074 psi) and a reinforcing steel stress σs0 = nsσc0 = -80 MPa (11.6 ksi).

Figure 7(b) illustrates the stresses and displacements at a section x under the action of a prestressing force P = P0 + δP at the stressing end. Between the concrete and the prestressing tendon, a bond shear force τb × pb per unit length is transferred, where τb is the average bond shear stress and pb is the bond perimeter.16 From equilibrium, one gets dσc = τb × pbdx/[Ac – Ad + (ns – 1)As]. Dividing both sides of this equation by Ec and considering that Ec[Ac – Ad + (ns – 1)As] = -P/εc1, where εc1 is the concrete strain in the unbonded length, leads to τb × pb = -(dεc/dx)(P/εc1) or, if strain increments δεc due to δP are considered, to

… (1)

Thus, at any section x, the bond shear force per unit length is proportional to the slope of the δεc-x curve and can be determined from experimental curves such as those shown in Fig. 6; the concrete strain increment δεc1 in the unbonded length is measured for each δP; hence, the proportionality factor on the right-hand side of Eq. (1) is known.

Whereas the displacements in the x-direction of the concrete uc and the reinforcing steel us are the same, there is a slip δ = up – uc between the concrete and the prestressing steel. Considering that the strain in the prestressing steel equals εp = σp/Ep = dup/dx and that τb × pb = -Ap(dσp/dx) one obtains, using Eq. (1)

… (2)

where δεp1 is the strain increment in the unbonded length of prestressing steel. Again, the second factor on the right hand side of Eq. (2) is a proportionality factor that can be derived for each δP from the measured strain increments δεc1 and δεp1 = δP/(ApEp), where Ep = 195 GPa (28,282 ksi)2 and Ap = 7 × 150 = 1050 mm2 (1.63 in.2) is the cross-sectional area of prestressing steel. Considering Eq. (2), it can be seen that the slip δ is proportional to the integral of δεc, extended from the section x to the end of the activated bond length.

In principle, Eq. (1) and (2) can be used to derive experimentally-based τb-δ relationships. The slip δ is proportional to the integral of δεc and can be determined in a fairly reliable way. The bond shear stress τb is proportional to the derivative of the small strain increments δεc, which can not be determined reliably.

Alternatively, one can start from idealized τb-δ relationships such as the trilinear relationship OABC shown in Fig. 8(a). In the δεc-x diagram shown in Fig. 8(b), the lines OA, AB, and BC of Fig. 8(a) correspond to exponential, linear, and harmonic curves, respectively. In A and B in Fig. 8(b), the exponential and harmonic curves have the same slope as the straight line AB. Starting from the coordinates of A, B, and C in Fig. 8(b), the characteristic values of Fig. 8(a) can be calculated, that is

… (3)

… (4)

… (5)

… (6)

… (7)


… (8)

λ follows from

… (9)

and pb can be determined as the smallest convex envelope of a strand bundle,16 that is

… (10)

where m is the number of seven-wire strands per tendon. Note that Eq. (6) and (7) are valid if .

As an example, consider the case δP = 467 kN (105.0 kips) shown in Fig. 6(a) and set xA = 1.82 m (71.65 in.), δεcA = -24 µε, xB = 1.58 m (62.20 in.), and δεcB = -50 µε; these coordinates of Points A and B were determined from a curve fitting process by applying the least squares method. With m = 7 and Ap = 1050 mm2 (1.63 in.2), Eq. (10) results in pb = 143.3 mm (5.64 in.). From Eq. (8), one gets κ = 9.231 km (5.736 miles). With δεc1 = -176 µε, Eq. (3) yields τb0 = 2.01 MPa (291 psi), and Eq. (9) results in λ = 1.670 m (65.8 in.); it can be seen that

where Ep = 195 GPa (28,282 ksi), Eq. (4) to (7) result in δA = 0.074 mm (0.003 in.), δB = 0.198 mm (0.008 in.), τbC = 0.724 MPa (105 psi), and δC = 2.895 mm (0.114 in.) (refer to Fig. 8(c)). Figure 8(d) compares the corresponding theoretical curve with the measured values of Fig. 6(a).

If there were a unique τb-δ relationship, the curves in Fig. 6 would be affine, that is, they should coincide if appropriately shifted in the x-direction. Obviously, this is not the case. If the procedure applied to the case δP = 467 kN (105.0 kips) of Fig. 6(a) were used for the other load stages, different τb-δ relationships would be obtained or, if the theoretical curve of Fig. 8(d) were shifted in the negative x-direction to match the cases δP = 299 kN (67.2 kips) and δP = 150 kN (33.7 kips), a considerable deviation between theory and experiment would be found. The curves in the activated bond length of Fig. 6(b) show a decreasing slope with increasing x, indicating that the softening branch of the τb-δ relationship was not reached; they show a better affinity than those of Fig. 6(a), yet they confirm that a unique τb-δ relationship is only acceptable as a rough approximation.

In summary, the bond tests with long embedment lengths resulted in concrete strain profiles with derivatives and integrals proportional to the bond shear stresses and slips, respectively. Whereas the untreated reference strands showed no softening bond shear stress-slip behavior, there was considerable softening for the emulsifiable oil-treated strands. A unique bond shear stress-slip relationship does not appear to be capable of describing the bond behavior over the entire activated bond length for variable pullout forces. As a rough approximation, average bond shear stresses over the entire activated bond length as found from the tests with the highest δP can be used, with activated bond lengths of approximately 2.25 and 0.9 m (88.6 and 35.4 in.), bond shear forces of 467 and 463 kN (105.0 and 104.1 kips), and a bond perimeter of pb = 143.3 mm (5.64 in.). This leads to average bond shear stresses of τb = 1.45 and 3.59 MPa (200 and 500 psi) for the treated and untreated strands, respectively, corresponding to a bond supply τb*pb of approximately 200 and 500 kN/m (13.7 and 34.3 kips/ft) for the treated and untreated strands, respectively.

Bond demand

Considering the bond reduction due to the emulsifiable oil treatment, the question arises whether this reduction is acceptable in practice. So far, this question has been answered negatively,11 and it is usually required that the emulsifiable oil is removed by flushing the ducts before grouting. This represents a significant drawback of an otherwise excellent temporary corrosion protection method. Actually, flushing of the tendons was proven to be ineffective,19 may leave water pockets in the tendons, and it is questionable whether the oil removal is justified by the bond demand.

It is not easy to make generally valid statements on the bond demand of post-tensioning tendons. However, it is quite easy to judge the bond demand for individual applications.

As an example, consider the simply supported beam shown in Fig. 9(a). The beam has a rectangular cross section with a width b of 400 mm (15.7 in.) and a depth h of 1 m (39.4 in.); its Span l equals 20 m (65.6 ft). The beam is assumed to be reinforced with only one parabolically draped tendon consisting of seven 15.7 mm (0.62 in.) diameter strands, that is, Ap = 1050 mm2 (1.63 in.2), post-tensioned to 1115 kN (250.7 kips) after long-term losses. For the sake of simplicity, friction losses are neglected. With its midspan eccentricity of 400 mm (15.7 in.), the tendon produces deviation forces of 1115 × 0.4/(202/8) = 8.92 kN/m (0.611 kips/ft) that almost compensate the beam’s selfweight g = 0.4 × 1.0 × 25 = 10 kN/m (0.685 kips/ft), assuming a specific weight of 25 kN/m3 (4.3 kips/yd3).

Figures 9(b) and (c) illustrate the beam’s behavior under the action of g and additional uniformly distributed loads q of 6 and 15 kN/m (0.41 and 1.03 kips/ft), respectively. Concrete and prestressing steel have been assumed to be linearly elastic, with the moduli Ec = 36.6 GPa (5308 ksi) and Ep = 195 GPa (28,282 ksi), respectively. In computing the cross-sectional moment-curvature relationships (Appendix A*), the concrete tensile strength was neglected. The shaded areas in the two figures correspond to the compressed zones with a depth z. The force flow can be recognized from the line of action of the compressive stress resultant and the tendon profile. Curvatures χ, deflections w, and tendon forces T are indicated for a number of cross sections.

Figure 9(b) corresponds to a typical service load behavior. There is some decompression of the concrete and the tendon force increase is small. The bond shear force dT/dx per unit length reaches a maximum of approximately 9.5 kN/m (0.65 kips/ft) between x = 4 and 6 m (157.5 and 236.2 in.).

Figure 9(c) corresponds to an overload situation near failure of the beam. The maximum tendon force of 1542 kN (346.7 kips) at midspan is equal to 83% of the characteristic tendon strength of 1859 kN (417.9 kips). The concrete is mainly decompressed and the maximum bond shear force per unit length amounts to approximately 81.6 kN/m (5.59 kips/ft) between x = 4 and 6 m (157.5 and 236.2 in.). Comparing this bond demand with a bond supply of approximately 200 kN/m (13.7 kips/ft) under very adverse conditions (wet strands), refer to Fig. 6(a), it can be seen that bond problems are unlikely to impair the beam’s load-deformation behavior even under the applied overload.

If the tendon had no bond at all, it would be strained and stressed uniformly over its entire length. Assuming, as a first approximation, a single vertical crack at midspan and considering the concrete to be rigid, a midspan deflection of 148.6 mm (5.85 in.) is obtained for g + q = 25 kN/m (1.71 kips/ft). The rotation of 148.6/10,000 = 14.9 mrad corresponds to an average strain of 14.9 × 0.9/10 = 1.337 mε, hence, to a tendon force of 1115 + 1.337 × 1050 × 195/1000 = 1389 kN (312.2 kips), corresponding to a bending moment of 1389 × 0.9 = 25 × 202/8 = 1250 kNm (922 kips-ft). A refined analysis (Appendix B*) results in a midspan deflection of 189.7 mm (7.5 in.), that is, 1.66 times the value of 114.0 mm (4.5 in.) given in Fig. 9(c).

Figure 10 compares the load-midspan deflection behavior for the fully bonded and the completely unbonded cases, starting from prestressing (Point A), over decompression (Point B) and service load (Point C), up to the considered overload (Point D).

Considering that the moment acting at any section of a member subjected to flexure and shear can be expressed as M = Tdv, where dv equals the lever arm of the internal forces and T equals the tensile force, and that the shear force V equals dM/dx, the bond shear force per unit length can be expressed as

… (11)

For a constant dv, dT/dx is proportional to V; and if there is no bond, T is constant and dv is proportional to M. Generally, as exemplified by Fig. 9(b) and (c) as well as Fig. 10, the behavior lies between these two limiting cases and the bond demand can be estimated by applying Eq. (11).


1. In the laboratory tests, a particular emulsifiable oil product used as a 25% aqueous emulsion showed the best corrosion protection behavior (refer to Fig. 1). In the field tests (Fig. 2), this product showed the best temporary corrosion protection behavior, independent of the duct (steel or plastic) and the prestressing steel (wires or strands) (refer to Fig. 3 and 4);

2. Under laboratory conditions, the surfaces of the emulsifiable oil-treated steels turned dry and waxy within a few days. Under field conditions, they remained wet and oily even after 6 months;

3. The development of a suitable quality control procedure for the application of emulsifiable oils is highly desirable;

4. Despite not being nearly as effective as the emulsifiable oil treatment, the dry air method can still be recommended for temporary corrosion protection. The inert gas method cannot be recommended, however;

5. Pullout tests with long embedment lengths are suitable for investigating the bond behavior of post-tensioning tendons (refer to Fig. 5);

6. No unique bond shear stress-slip relationship appears to be capable of describing the bond behavior over the entire activated bond length for variable pullout forces (refer to Fig. 6 and 8). The actual, three-dimensional bond behavior is merely approximated by a one-dimensional model (Fig. 7) involving a unique bond shear stress-slip relationship;

7. As a rough approximation, average bond shear stresses over the entire activated bond length can be used for comparing tests as well as for practical applications. On this basis, emulsifiable oil-treated strands with wet surfaces showed a bond shear stress reduction by a factor of approximately 2.5 compared with untreated strands (refer to Fig. 6);

8. The bond demand can be estimated by applying standard methods of structural analysis and prestressed concrete theory (Appendix A). Under service loads, the bond demand is typically found to be a small fraction of the bond supplied by emulsifiable oil-treated strands with wet surfaces and even near ultimate loads it is unlikely that the bond demand will exceed the supply (refer to Fig. 9);

9. Based on a detailed example analysis and similar considerations for other applications, it can be concluded that generally, because of bond reduction, emulsifiable oils applied for temporary corrosion protection of post-tensioning steel do not need to be removed before grouting the tendons. The application of emulsifiable oils, however, is not permissible for bonded dead-end anchorages; and

10. The bond reduction caused by emulsifiable oils in grouted tendons may lead to a somewhat softer-than-usual loaddeformation response of post-tensioned concrete members in the decompressed state. If deemed to be necessary, a lowerbound estimate of the corresponding stiffness can be obtained by assuming a completely unbonded behavior (refer to Fig. 10) of the post-tensioning tendons (Appendix B).


The financial support from the Swiss Federal Roads Authority (ASTRA) is gratefully acknowledged.


Ac = cross-sectional area of gross cross section

Ad = cross-sectional area of duct

Ap = cross-sectional area of prestressing steel

As = cross-sectional area of reinforcing steel

b = width

dv = lever arm of internal forces

Ec = modulus of elasticity of concrete

Ep = modulus of elasticity of prestressing steel

Es = modulus of elasticity of reinforcing steel

e = eccentricity of tendon

g = dead load per unit length

h = depth

Ic = moment of inertia of gross cross section

l = span

M = bending moment

Mdec = decompression moment

m = number of seven-wire strands

np = modular ratio (Ep/Ec)

ns = modular ratio (Es/Ec)

P = prestressing force

P0 = initial prestressing force

P1 = force in unbonded tendon at decompression

P2 = force in unbonded tendon in decompressed state

pb = bond perimeter (smallest convex envelope of strand bundle)

q = live load per unit length

q1 = live load at decompression (unbonded system)

q2 = live load in decompressed state (unbonded system)

qdec = live load at decompression (bonded system)

T = resultant internal tensile force

uc = concrete displacement in x-direction

uc0 = concrete displacement in x-direction due to P0

up = prestressing steel displacement in x-direction

us = reinforcing steel displacement in x-direction

V = shear force

w = deflection

wm = midspan deflection

wm0 = midspan deflection due to P0

wm1 = midspan deflection at decompression (unbonded system)

wm2 = midspan deflection in decompressed state (unbonded system)

x = coordinate

z = depth of compression zone

α = e/h at midspan

β = related midspan eccentricity of compressive stress resultant

χ = curvature

χ0 = curvature at midspan

δP = prestressing force increment

δεc = concrete strain increment due to δP

δεc1 = concrete strain increment in unbonded length

δεp1 = prestressing steel strain increment in unbonded length

δ = slip between prestressing steel and concrete (up – uc)

εc1 = concrete strain in unbonded length

εcp = concrete strain at depth of tendon

εcp0 = average εcp due to P0

εcp1 = average εcp at decompression

εcp2 = average εcp in decompressed state

εp = prestressing steel strain

κ = length parameter

λ = length parameter

ρp = prestressed reinforcement ratio

ξ = 1 – 2x/l

ξ1 = related length of decompressed region

σc = concrete stress

σc0 = concrete stress due to P0

σp = prestressing steel stress

σp0 = prestressing steel stress due to P0

σs = reinforcing steel stress

σs0 = reinforcing steel stress due to P0

τb = bond shear stress

τb0 = peak value of τb

* The Appendix is available at www.concrete.org in PDF format as an addendum to the published paper. It is also available in hard copy from ACI headquarters for a fee equal to the cost of reproduction plus handling at the time of the request.


1. Fuzier, J.-P.; Ganz, H. R.; and Matt, P., “Durability of Post-Tensioning Tendons,” Recommendation, Bulletin 33, Fédération Internationale du Béton (fib), 2006, 71 pp.

2. Swiss Society of Engineers and Architects, “Concrete Structures,” SIA 262, Zurich, Switzerland, 2004, 90 pp.

3. Federal Department for Environment, Traffic, Energy and Communication/ Federal Roads Authority/SBB Ltd., “Measures to Ensure the Durability of Post- Tensioning Tendons in Bridges,” Guideline No. 308.322e, Berne, Switzerland, 2001, 14 pp.

4. Isecke, B., and Stichel, W., “Einfluss baupraktischer Umgebungsbedingungen auf das Korrosionsverhalten von Spannstaehlen vor dem Injizieren (Influence of Environmental Conditions on the Corrosion Behavior of Prestressing Steel Prior to Grouting),” Research Report No. 87, Federal Institute for Materials Research and Testing (BAM), Berlin, Germany, 1982, 49 pp.

5. Salcedo-Rueda, E.; Schokker, A. J.; Breen, J. E.; and Kreger, M. E., “Bond and Corrosion Studies of Emulsifiable Oils Used for Corrosion Protection in Post-Tensioned Tendons,” PTI Journal, V. 1, No. 3, 2004, pp. 30-38.

6. Rieche, G., and Delille, J., “Erfahrungen bei der Pruefung von temporaeren Korrosionsschutzmitteln fuer Spannstaehle (Experiences from Testing of Temporary Corrosion Protection Agents for Prestressing Steel),” Deutscher Ausschuss fuer Stahlbeton, V. 298, Berlin, Germany, 1978, pp. 1-19.

7. Werner, R.; Faller, M.; Richner, P.; and Matt, P., “TEKORS-Temporaerer Korrosionsschutz von Spanngliedern-Wirksamkeit und Praxistauglichkeit (TEKORS-Temporary Corrosion Protection of Prestressing Tendons- Effectiveness and Practical Suitability),” Research Report No. 14.01, Laboratory for Corrosion and Materials Integrity, Empa, Dubendorf, Switzerland, Dec. 2004, 54 pp.

8. Isecke, B.; Mietz, J.; and Schuett, K., “Temporaerer Korrosionsschutz von Spannstaehlen in unverpressten Huellrohren (Temporary Corrosion Protection of Prestressing Steels in Non-Injected Ducts),” Materials and Corrosion, V. 54, 2003, pp. 413-418.

9. Danzer, W., and Rebhan, D., “Spannstahlkonservierung mit Stickstoff (Preservation of Prestressed Steels Using Nitrogen),” Betonwerk + Fertigteil-Technik, No. 2, 1981, pp. 87-90.

10. Helminger, E., and Ruhl, K., “Temporaerer Korrosionsschutz von gespannten Spannstaehlen in Spannbetonbauteilen unter Verwendung von Stickstoff (Temporary Corrosion Protection of Prestressed Steel in Prestressed Concrete Members Using Nitrogen),” Bauingenieur, V. 56, 1981, pp. 395-399.

11. Isecke, B.; Mahlcke, W.; Mietz, J.; and Rueckert, J., “Temporaerer Korrosionsschutz von Spannstaehlen mit filmbildenden Mitteln (Temporary Corrosion Protection of Prestressing Steels with Film Forming Coatings),” Materials and Corrosion, V. 48, 1997, pp. 613-623.

12. DIN 51386-1, “Korrosionsschutzoele-Prüfung im Kondenswasser- Wechselklima (Corrosion Preventive Oils-Testing in Condensation Water Alternating Atmosphere),” German Institute of Standardization, 1982, 5 pp.

13. DIN 50017, “Klimate und ihre technische Anwendung-Kondenswasser- Prüfklimate (Atmospheres and Their Technical Application- Condensation Water Test Atmospheres),” German Institute of Standardization, 1982, 5 pp.

14. ASTM G31-72, “Standard Practice for Laboratory Immersion Corrosion Testing of Metals,” ASTM International, West Conshohocken, PA, 2004, pp. 98-105.

15. “VSL Construction Systems 2000,” http://www.vsl-intl.com, 32 pp.

16. Marti, P., “Verbundverhalten von Spanngliedern mit Kunststoff- Huellrohren (Bond Behavior of Prestressing Tendons with Plastic Ducts),” SP-001, Institute of Structural Engineering, ETH, Zurich, Switzerland, 1994, pp. 143-150.

17. Rogowsky, D. M., and Marti, P., “Detailing for Post-Tensioning,” VSL Report Series No. 3, VSL International Ltd., Berne, Switzerland, 1991, 49 pp.

18. EN 197-1, “Cement-Part 1: Composition, Specifications and Conformity Criteria for Common Cements,” European Committee for Standardization, Dec. 2000, 26 pp.

19. Davis, R. T.; Tran, T. T.; Breen, J. E.; and Frank, K. H., “Reducing Friction Losses in Monolithic and Segmental Bridge Tendons,” Research Report No. 1264-2, Center for Transportation Research, Bureau of Engineering Research, University of Texas at Austin, Austin, TX, Oct. 1993, 118 pp.

ACI member Peter Marti is a Professor of structural engineering and Head of the Department of Civil, Environmental and Geomatic Engineering at the Swiss Federal Institute of Technology (ETH), Zurich, Switzerland. His research interests include structural concrete and masonry.

Robert Ullner is a Research Associate at ETH and the Swiss Federal Laboratories for Materials Testing and Research (Empa), Dubendorf, Switzerland. His research interests include prestressed concrete and bond between concrete and reinforcement.

Markus Faller is Deputy of the Laboratory for Corrosion and Materials Integrity at Empa. His research interests include the corrosion protection of prestressing steel, the use of metals in indoor swimming pools and tunnels, and corrosion damage of metals for the construction and machine building industry.

Christoph Czaderski is a Research Engineer and Group Leader at the Structural Engineering Research Laboratory of Empa. His research interests include the strengthening of reinforced concrete structures using fiber reinforced polymers and the application of shape memory alloys for civil structures.

Masoud Motavalli is Head of the Structural Engineering Research Laboratory at Empa and an Assistant Professor at the University of Tehran, Tehran, Iran. His research interests include the application of advanced materials such as polymer composites, shape memory alloys and adaptive systems in structural engineering, structural post-strengthening, and seismic upgrading of existing structures.

Copyright American Concrete Institute Jan/Feb 2008

Provided by ProQuest Information and Learning Company. All rights Reserved