Causes of Deterioration of Precast Bridge Piles: An Experimental Study Ahmad Shayan, Chief Scientist, ARRB Group Ltd. Ahmad Shayan is a Chief Research Scientist at ARRB Group, Melbourne, Australia, and is responsible for the Concrete Group. His research interests include concrete materials durability, deterioration problems in concrete structures, utilisation of industrial waste materials in concrete, and steam-cured concrete products. He is a member of several national and international committees, and won the 2003 Clunies Ross medal for his work on durability of concrete structures. Contacts:

Phone: Fax: Email:

03 9881 1658 03 9803 2611 [email protected]

Aimin Xu, Senior Research Engineer, ARRB Group Ltd. Aimin Xu is a Senior Research Engineer in the Concrete Group. Prior to starting his career in Australia, Aimin worked within the FORCE Institute, Denmark which specialises in non-destructive testing of concrete structures, and SAQ Kontroll AB, another non-destructive testing company in Sweden. His research interests include chemistry of cement and concrete, non-destructive testing procedures, including electrochemical techniques for corrosion measurements and corrosion diagnosis and assessment of structures. Contacts:

Phone: Fax: Email:

03 9881 1530 03 9803 2611 [email protected] .

Adrian Hii, Research Engineer, ARRB Group Ltd. Adrian Hii is a postgraduate research student in structural engineering at Monash University. His research interests include the rehabilitation of reinforced concrete structures with externally-bonded FRP. He is assigned to ARRB Group as a Research Engineer on a part time basis. Contacts:

Phone: Fax: Emails:

03 9881 1637 03 9803 2611 [email protected]

Causes of deterioration of precast bridge piles: An experimental study Ahmad Shayan* - Chief Scientist, Aimin Xu - Senior Research Engineer, and Adrian Hii - Research Engineer, ARRB Group Ltd

Synopsis Recently several cases of deterioration have been noted in precast bridge piles located in marine environments. The deterioration has been attributed to alkaliaggregate reaction (AAR), delayed ettringite formation (DEF), and salt attack on concrete, or a combination of some of these factors. Moreover, similar cases of deterioration have also been noted in non-marine conditions. To clarify the influence of the different factors, 16 precast and cast-in-situ, reinforced model piles were manufactured, which incorporated different types of aggregates and sulfate contents, at ambient temperature (20°C) or by steam curing at 85°C. Strain measurements were made on both concrete and reinforcement bars under marine and non-marine exposure conditions. Significant expansion associated with cracking has occurred in the piles containing the reactive aggregate, and the expansion appears to be larger in the bottom, wet areas of piles than in the top, drier areas. Larger expansions were noted where additional sulfate was present to cause DEF in the steam cured piles containing reactive aggregates. Representative piles have been repaired by epoxy-bonded, fibre-reinforced polymer (FRP) wrapping, and concrete jacketing, and are being monitored, through extended strain measurement, to evaluate the effectiveness of the repair.

1.

Introduction

Recently several serious cases of deterioration of precast piles have been noted in bridges in tidal water as well as fresh water. In some cases the concrete piles had lost a portion of the concrete and the reinforcement which caused serious concern over the load capacity of the structure concerned. The causes of the deterioration appear to be multiple and can be attributed to alkaliaggregate reaction (AAR), corrosion of steel reinforcement, internal sulfate attack (delayed ettringite formation - DEF) and salt attack from the environment. It is very important to identify the primary causes of the deterioration, because if these are prevented, then the secondary causes could be arrested as well. AAR is well known to be able to cause serious cracking in the affected concrete. This exposes the interior of the concrete to attack by aggressive agents such as oxygen, carbon dioxide, moisture and salt water. Subsequent to AAR damage, salt attack on concrete and corrosion of reinforcement can also proceed at a much faster rate. Therefore, these problems could be minimised if the concrete is free from AAR, which can be achieved by correct selection of the aggregate phase.

Delayed ettringite formation (DEF) is related to cement composition and heat-curing temperature, applied to precast elements. DEF allegedly causes expansion and deterioration of the cement matrix in the concrete on its own right. However, a case of DEF alone has not been documented in Australia, where symptoms of DEF have always been associated with AAR damage. If DEF is actually able to cause expansion and deterioration independently of AAR, then there would be more cause for concern. The volumetric changes in concrete, induced by AAR and DEF, cause cracking of concrete and generate stresses in the reinforcement which could exceed its yield strength. Prevention of DEF damage is different from that of AAR damage, and involves control of cement composition, such as sulfate, aluminate and alkali contents. It is also important to find out whether the heat-cured concrete elements are more prone to damage than normally cured concrete. Uninformed repair of damaged elements subjected to on-going expansion (such as AAR, DEF or corrosion) could be a waste of money. Unfortunately, the effectiveness of repair systems such as carbon fibre composite wrapping and concrete jacketing has not been documented for such deteriorating concrete, and there is a great need to develop knowledge in this area. Research is needed to generate sound data on the performance of materials and manufacturing processes with respect to AAR and DEF and on the effectiveness of appropriate repair systems to increase the service life of damaged elements. This would enable durable structures to be specified with respect to AAR and DEF, and existing damaged elements repaired effectively.

2.

Scope of the study

This research project consisted of two stages. The objective of the first stage was to manufacture model reinforced concrete piles and to determine the extent of strain development in the concrete and in the reinforcement bars, and monitor cracking in the different piles. The objective of the second stage was to repair representative cracked piles by concrete jacketing and carbon fibre reinforced polymer (CFRP) wrapping, and to compare the effectiveness of the two repair methods in restraining the AAR expansion. Figures (1A &1B) show the combinations of the variables for stage 1, and Table 1 gives the levels of the various variables. Table 1 - Variables used for casting of experimental piles

Factor

No. of variables

Levels

Aggregate reactivity

3

high (Hr), low (Lr), and non-reactive (Nr).

Initial Curing

2

Normal curing, 20°C (N); Steam-curing, 85°C (S)

Cement SO3 content

2

2.6 % (native content ), 5 % (by added gypsum)

Exposure

3

Atmospheric (A), tap water (W), 3.5% salt water (C)

Causes of deterioration of precast bridge piles: An experimental study, Shayan

2

Lr

Nr

Aggregate type N

Initial curing W

Exposure

N

S

C

W

Hr

W

C

S

C

W

N

C

W

S

C

W

C

Figure 1A – Combination of variables in the first set of 12 piles. These are designated piles 1-12 from the left to the right.

Aggregate type

Hr + Sulfate

Nr + Sulfate

Initial curing

S

S

N

S

Exposure

W

C

W

W

Figure 1B - Combination of variables in the second set of four piles. These are designated piles 13-16 from the left to the right.

Model piles were manufactured as detailed below, using the various combinations shown in Figures 1A and 1B. The specimens were labelled in the order of the aggregate types used (Nr, Lr or Hr), curing conditions (N or S) and the environmental exposure (W or C). For example, ‘Hr-S-C’ indicates highly reactive aggregate, steam cured pile and half-immersion in salt water. The top portion of the half-immersed specimens provides the atmospheric exposure condition. The General Purpose (GP) cement used had an alkali content of 0.58% Na2O equivalent and a C3A (tricalcium aluminate) content of about 5%. The concrete mix selected contained 450 kg cement /m3 with a cement alkali level of 1.4 %, i.e., a total concrete alkali content of 6.3 kg/m3. Curing under normal temperature would generate AAR expansion alone for reactive aggregate, whereas curing at 85°C could generate both AAR and DEF with reactive aggregate, and DEF alone with nonreactive aggregate. The original native sulfate content of the cement (2.6 % SO3) was used in the first set of 12 piles. This was raised to 5% by cement mass in the second set of four piles to induce DEF under steam curing conditions. The effects of exposure to salt water can be clarified by comparing undamaged concrete to those damaged by AAR and/or DEF mechanisms. This paper deals with stage 1 of the project. Repairing of the damaged piles has recently been completed and the results will be published later when adequate data has been collected.

Causes of deterioration of precast bridge piles: An experimental study, Shayan

3

3.

Specimen configuration

Figure 2 shows the cross sectional details of the 1100 mm long model piles, and Figure 3 the longitudinal section and the positions of the various strain gauges. 300 Concrete cover of 50 mm R10 stirrup bars at 250mm

300

4Y20 bars Reinforcement ratio 1.4% 160 Figure 2 – Cross section and reinforcement details of pile specimen

In piles 10, 11, and 12 fibre optic sensors specific to both concrete and steel were placed in the top portion of the pile, i.e. atmospheric zone, as well as in the submerged bottom portion, for comparison with the other sensors installed in the piles.

Surface mounted strain gauges on steel reinforcement 550

Strain gauges embedded within concrete measuring lateral expansion Thermocouples Weldable steel strain gauges on steel reinforcement

Half-immersed in water or salt solution 550

Figure 3 – Longitudinal section of model piles and positions of strain gauges

Causes of deterioration of precast bridge piles: An experimental study, Shayan

4

Table 2 lists the various types of strain gauges used, and Table 3 provides the designations employed for the gauges in the various locations. Table 2 - Details of strain gauges used

Gauge length (mm)

Strain Gauge type CEA-06-250UN-120 CEA-06-W250A-120 N11-FA-60-120-11 EGP-5-120 FISO fibre optic FISO embedded fibre optic

6 6 60 120 N/A 60

Description Steel surface strain gauge Steel weldable strain gauge Concrete surface strain gauge Concrete embedded strain gauge Steel surface strain gauge Concrete embedded strain gauge

Table 3 - Strain gauge designation

TM2W TM4S TS2, TS4 BS4 BM2S, BM4S BS2 TCC, TCS BCC, BCS TC3, TC4 BC3, BC4 C#TS C#TC C#BS C#BC

Top half, Main bar, 2nd bar, Welded strain gauge Top half, Main bar, 4th bar, Surface strain gauge Top half, Shear reinforcement, 2nd leg and 4th leg Bottom half, Shear reinforcement, 4th leg Bottom half, Main bar, 2nd bar, and 4th bar surface strain gauge Bottom half, Shear reinforcement, 2nd leg Top half, Concrete embedded strain gauge, and centre side of beam, respectively Bottom half, Concrete embedded strain gauge, and centre side of beam, respectively Top half, Concrete surface strain gauge, 3rd and 4th surface Bottom half, Concrete surface strain gauge, 3rd and 4th surface Pile #, Top half, Steel type fibre optic sensor Pile #, Top half, Concrete type fibre optic sensor Pile #, Bottom half, Steel type fibre optic sensor Pile #, Bottom half, Concrete type fibre optic sensor

4. Concrete mix proportion, early age treatment and storage environment The cement content for all mixes was 450 kg/m3. The water to cement ratio was 0.42, and the coarse aggregate content depended on the aggregate density; being 1130, 1060 and 1100 kg/m3 for the Nr, Lr and Hr aggregates, respectively. The fine aggregate content was 690 kg/m3 for all mixes. At the time of casting the piles, accompanying concrete cylinders (100 mm diameter by 200 mm long) and concrete prisms (75 x 75 x 285 mm) were also cast for strength testing and expansion measurement, respectively.

Causes of deterioration of precast bridge piles: An experimental study, Shayan

5

For the steam-cured specimens, the preheating period after casting was 2.5 - 3.0 hours, after which they were placed in the steam chamber and the temperature was raised by 10°C/ hr until it reached 85°C. The soaking period at the maximum temperature was five hours for the first set of model piles, but 10 hours for the second set. After the steam curing period, the heating was automatically cut off and samples were allowed to cool down in the closed chamber to ambient temperature. After demoulding, all the piles were wrapped in moist cloth and, after one week, surface strain gauges were installed in appropriate locations. Subsequently, the model piles were wrapped in moist cloth again and then half-immersed in their containers, either in water or salt solution. In the case of salt solution, the moist wrapping was present only in the top part of the pile, so that salt does not migrate up the column. The whole assembly was then wrapped in plastic sheeting and placed at 38°C for expansion measurement. Figures 4 to 6 show the various stages of preparation of the model piles for expansion measurements.

Figure 4 - Steel cage

Figure 5 - Finished pile

5.

Results

5.1

Properties of concrete mixes

Figure 6 - Storage of pile

The results of 28-day compressive strength tests on concrete cylinders are shown in Figure 7. The steam-curing somewhat reduced the 28-day strength, and the reduction is larger for the reactive than the non-reactive aggregate. This means that steam curing probably induced some AAR expansion and micro-cracking in the concrete by the age of 28 days. The 1-year expansion results of concrete prisms stored at 38°C, 100% relative humidity (after the initial curing), are given in Table 4. Steam curing significantly increased the AAR expansion of concrete prism compared to normal curing, and this was more evident for the slowly reactive aggregate. The increased AAR expansion of the specimens containing the slowly-reactive aggregate

Causes of deterioration of precast bridge piles: An experimental study, Shayan

6

arose due to thermal activation of the reaction between the alkali and reactive silica (microcrystalline quartz). The large expansion of concrete prisms containing the highly reactive aggregate (Hr) under both normal and steam-curing conditions clearly demonstrates the high AAR-susceptibility of this aggregate. Table 4 - Expansion of concrete prisms Concrete

1-year Expansion (%)

Nr-N

0.010

Nr-S

0.003

Lr-N

0.032

Lr-S

0.178

Hr-N

0.140

Hr-S

0.166

Figure 7 - Compressive strength at 28 days

The smaller expansion value for the steam-cured specimens containing the highly reactive aggregate indicates that the thermal activation rate was high and part of the expansion took place during the steam-curing and was missed in the subsequent measurement. It should be noted that the AAR expansion induced in the reinforced piles could be much less than that exhibited by the concrete prisms which are not reinforced, as it is well known that steel reinforcement significantly reduces the AAR expansion (Hobbs, 1988). 5.2

Measurements on model piles

Strain measurements were made for gauges on the concrete surface, embedded within the concrete and on the steel bars. Nearly all surface gauges failed due to breakdown of their protective coating, which resulted from the very humid environment and moisture condensation on them. Reasonable readings were obtained from the other gauges, although some fibre optic gauges also failed soon after installation. 5.2.1

Concrete gauges

The strain readings from the concrete embedded gauges in normal- and steam-cured piles, containing the different aggregates and half-immersed in salt water, are compared in Figure 8. The piles half-immersed in water showed the same trend. At the present time exposure to salt water has not shown additional effects on expansion, but at later ages the reinforcement bars may suffer corrosion and the concrete matrix may show adverse effects. Pile expansion followed the reactivity of the aggregate in it. Steam curing increased the AAR expansion of the low-reactive aggregate, whereas the piles containing the highly reactive aggregate showed decreased expansion. The bottom portion of the pile which was immersed in water or salt solution expanded more than the top portion which was under drier condition.

Causes of deterioration of precast bridge piles: An experimental study, Shayan

7

250

200

200

150

150 Microstrain

Microstrain

250

100 50 0 -50

100 50 0

0

200

400

600

800

-50

-100

0

200

800

800

700

700

600 500

600 Microstrain

Microstrain

Steam curing, half-immersed in salt solution, Non-reactive aggregate

400 300 200 100 0

500 400 300 200 100

200

400

600

0 -100 0

800

-200

200

Age (days)

400

600

800

A ge (days)

Normal curing, half-immersed in salt solution, Low reactivity aggregate

Steam curing, half-immersed in salt solution, Low reactivity aggregate

4000

1600

3500

1400

3000

1200

2500

1000

Microstrain

Microstrain

800

Age (days)

Normal curing, half-immersed in salt solution, Non-reactive aggregate

2000 1500 1000 500

800 600 400 200

0 -500 0

600

-100 A ge (days)

-100 0

400

200

400

600

800

0 -200 0

200

A ge (days)

400

600

800

Age (days)

Normal curing, half-immersed in salt solution, highly reactive aggregate

Steam curing, half-immersed in salt solution, highly reactive aggregate

TCS - Concrete gauge on top half and at side of pile TCC - Concrete gauge on top half and at centre of pile BCS - Concrete gauge on bottom half and at side of pile BCC - Concrete gauge on bottom half and at centre of pile

Figure 8 - Concrete strain development in piles with the three different aggregates

Causes of deterioration of precast bridge piles: An experimental study, Shayan

8

The strain measurements (expansion) for piles 13 and 14 (containing the Hr aggregate), and piles 15 and 16 (containing the Nr aggregate) are shown in Figure 9. The two piles, with the Hr aggregate and additional sulfate (piles 13 and 14) expanded significantly in a relatively short period. Compared to similar piles without the additional sulfate, the expansion was larger for the same age, probably indicating that DEF exacerbated the AAR expansion. For piles 15 and 16, which incorporate the non-reactive aggregate, no AAR expansion is expected. However, due to the high alkali and sulfate contents, and the high steam-curing temperature reached (88°C, within the concrete), the conditions were very favourable for DEF formation. If DEF alone can cause expansion and cracking, then it should be observed in these piles. No conclusion can be made at the present time regarding DEF expansion, due to the young age of piles. Further monitoring will be done in the future to enable a sound conclusion to be made. 3000

3500

2500

3000 2500 Microstrain

Microstrain

2000 1500 1000 500

1500 1000 500 0

0 -500

2000

0

100

200

300

-500 0

400

100

600

250

500

200

400

150

300 200

100 50

100

0

0

-50

100

400

Steam curing, half-immersed in salt solution, Highly reactive aggregate, 5% SO3

Microstrain

Microstrain

Steam curing, half-immersed in water, Highly reactive aggregate, 5% SO3

0

300

A ge (days)

A ge (days)

-100

200

200

300

0

100

200

300

-100

A ge (days)

A ge (days)

Steam curing, half-immersed in water, Non-reactive aggregate

Normal curing, half-immersed in water, Non-reactive aggregate, 5% SO3

TCS - Concrete gauge on top half and at side of pile TCC - Concrete gauge on top half and at centre of pile BCS - Concrete gauge on bottom half and at side of pile BCC - Concrete gauge on bottom half and at centre of pile

Figure 9 - Concrete strain developments in columns with sulfate (columns 13 to 16)

Causes of deterioration of precast bridge piles: An experimental study, Shayan

9

5.2.2

Steel gauges

The steel gauges for the piles with non-reactive aggregate were erratic, and those with the slowly reactive aggregate sometimes recorded very large readings and appear to be incorrect. Examples of results for the main steel bars and stirrups in piles with the highly reactive aggregate are presented in Figures 10. Considering that the maximum concrete expansion has largely been less than 2000 microstrains at the time of the last measurement, steel strains of greater than this value may not be realistic. Sensors which measured negative strains may have developed faults due to high humidity. Ste e l gauge s on m ain bars

Ste e l gauge s on m ain bars

4000

3500

3000

3000 Microstrain

Microstrain

2500

2000 1000 0 -1000

0

100

200

300

400

2000 1500 1000 500

500

0 -500 0

-2000

100

300

400

Normal curing, half-immersed in salt solution, Main bar

Steam curing, half-immersed in salt solution, Main bar

Ste e l gauge s on s tirrups

Ste e l gauge s on s tirrups

2500

1500

2000

1000

1500 1000 500

500

500 0 0

100

200

300

400

500

-500

0 -500

200

A ge (days)

Microstrain

Microstrain

A ge (days)

0

100

200

300

400

500

-1000 A ge (days)

Age (days)

Normal curing, half-immersed in salt solution, Stirrup

Steam curing, half-immersed in salt solution, Stirrup

TM2W - Main bar gauge on top half and at 2nd side of pile TM4S - Main bar gauge on top half and 4th side of pile BM2S - Main bar gauge on bottom half and 2nd side of pile BM4S - Main bar gauge on bottom half and 4th side of pile

Figure 10 - Strain development in the main bar and stirrup of piles made with highly reactive aggregate

Causes of deterioration of precast bridge piles: An experimental study, Shayan

10

5.3

Comparison of fibre optic and electrical sensors

Results presented in Figures 11 and 12 show that there are certain variations between the readings from the fibre optic gauges and the foil gauges. The trends are, however, similar. For the concrete gauges (triangle and diamond in Figure 11), more expansion was recorded in the lower wet part of the pile (1000 microstrains) than in the top part (650 microstrains). The fibre optic gauge in the wet area (x) appears to have become faulty after one year. The readings of the two sensor types were very similar for the top part of the pile. 2000 Top of pile, f ibre optic

Microstrain

1500

Top of pile 1000 500

Bottom of pile, f ibre optic

0

Bottom of pile 0

200

400

600

800

-500 A ge (days)

Figure 11 - Comparison of readings from the concrete type of fibre optic and electrical embedded gauges in pile 12 (Fibre optic gauges: x and „) T and B in the designation refer to the top and bottom portions of the pile.

The normal steel type of sensors (electrical) showed some erratic behaviour, but the final measurements were similar for the top and bottom portions of piles. The fibre optic type of steel gauge showed much lower strains for the top portion of pile. The gauge in the lower, wet portion of the pile appears to have failed after about 300 days. Ste e l gauge s on s tirrups 2500 Top of pile

Microstrain

2000 1500

Bottom of pile

1000 Top of pile, f ibre optic

500 0 -500 0

200

400

600

800

Bottom of pile, f ibre optic

-1000 A ge (days)

Figure 12 - Comparison of readings from the steel type of fibre optic (hollow symbols) sensor and electrical (solid symbols), embedded gauges in pile 12

Causes of deterioration of precast bridge piles: An experimental study, Shayan

11

5.4

Repair of cracked piles

Piles 10, 11, 13 and 14, which had exhibited sufficient cracking, were selected for repair. Piles 10 and 13 were repaired using conventional concrete jacketing, and piles 11 and 14, which had more cracking by the CFRP. Figures 13 and 14 show the extent of cracking on the various faces of piles 11 and 14, respectively. The repair was carried out only a month ago, and sufficient data in not available at this stage, but the results will be reported elsewhere when available.

Figure 13 - Cracking developed in pile 11

6.

Figure 14 – Cracking developed in pile 14

Conclusions

Measurements made so far on the model piles, which contained a sufficient amount of alkali, have shown that : •

The reactivity of the aggregate plays a major role in the deterioration.

•

Steam-curing at 85°C, per se, did not cause any deterioration in the concrete containing non-reactive aggregate, when the native SO3 and C3A contents of the cement were not increased.

•

Steam-curing increased the reactivity of the slowly reactive aggregate. The measured expansion of steam-cured concrete containing highly reactive aggregate was lower than that of normal concrete, either because some reaction took place during the steam-curing process, or because of the larger porosity of steam-cured concrete, or both.

•

Expansion was greater in the submerged zone of the piles.

•

Increased sulfate content of cement caused larger expansion in the presence of reactive aggregate, probably because DEF enhanced the AAR expansion.

Causes of deterioration of precast bridge piles: An experimental study, Shayan

12

•

Surface-mounted strain gauges were unsuitable for expansion measurement, whereas embedded concrete gauges were satisfactory. Steel gauges and fibre optic gauges did not perform satisfactorily in wet areas and developed faults.

Measurements are continuing and further interpretation will be made when adequate data is available on the performance of the repair systems.

7.

Acknowledgments

The Authors express their gratitude to Main Roads Queensland, Main Roads WA, RTA NSW, and VicRoads for funding this project.

8.

Reference

Hobbs, D.W. (1988). “Alkali silica reaction in concrete.” Thomas Telford, London, UK.

Causes of deterioration of precast bridge piles: An experimental study, Shayan

13