Keywords

1 Introduction

Patch repair of concrete is a common requirement for reinforced concrete structures. While it is understood that the fundamentals of best practices and performance criteria in concrete repair also apply to heritage concrete, the many studies and published guidance [1,2,3,4,5,6,7,8,9] on the performance criteria of concrete repairs are not always familiar to the conservation field. In some cases, they promote irreversible changes to the structure that are in conflict with traditional conservation principles such as minimal intervention and retreatability. While there is now a general acceptance of the importance of concrete heritage from the 20th century [10], the few widely accepted guidelines on the approach to its preservation and conservation [11,12,13,14,15] do not specifically address these issues or provide guidance on the long-term performance of patch repairs which were designed to preserve the aesthetic significance of the original fabric.

These issues were raised as key concerns during an international experts’ meeting focused on conserving concrete heritage which took place at the GCI, Los Angeles, in 2014 [16]. As a response to this, the Getty Conservation Institute (GCI), Historic England (HE) and the Laboratoire de Recherche des Monuments Historiques (LRMH) commenced work on an international collaborative research project, ‘Performance Evaluation of Patch Repairs on Historic Concrete Structures’ (PEPS), in 2018. The project team, which is also joined by expert consultants Rowan Technologies (England) and Wiss, Janney, Elstner Associates (USA), aims to produce practical guidance to help those repairing historic concrete through the process of the selection of appropriate repair approaches, procedures, and materials. Case studies have been assessed in the USA, England, and France, with a variety of climatic and environmental conditions, typologies, and repair materials. Previous publications have addressed the background and development of the project [16,17,18,19], the project methodology [20, 21], and preliminary results from each country [22,23,24].

The project is composed of four stages of work:

  • Phase I Project development

  • Phase II Preliminary assessment

  • Phase III Detailed diagnostic

  • Phase IV Synthesis & conclusions

Phases I-III have been completed at five case studies in England. This paper presents preliminary results from the Phase III in-situ assessments of two of these case studies, between which the approach to the repair methodology varied. The case studies are not publicly identified and, instead, have been allocated unique reference codes: ENG04, and ENG05.

2 Case Study I: ENG05

2.1 Background

ENG05 was built between 1935–1937 and is one of 6 structures on a private property which are listed at Grade II* or Grade II. The structure is subject to exposure classes XC4 (cyclic wet and dry) and XF1 (moderate water saturation, without de-icing salts) [25], and had a widespread issue of inadequate cover to the steel reinforcement combined with carbonation depths up to 40 mm. This had resulted in extensive areas of reinforcement corrosion and concrete spalling. A major repair and conservation project was carried out on five of the structures on the property between 2014 and 2016, during which the property owners established their own concrete repair department to carry out the repairs with their architects and engineers. The structure in the worst condition prior to repairs was selected for examination during the Phase III assessment. ‘Like-for-like’ repairs (same cement type with similar aggregate) were specified and trial mixes undertaken to determine appropriate mix designs. This approach was specified by the engineer in order to provide both a compatible material with similar properties which, in theory, would reduce the risk of future differential movements, and a closer visual match to the original fabric.

The standard repair for use on the top of level surfaces and vertical faces was a 1:2:4 (cement:sand:aggregate) mix to match the original concrete. A maximum aggregate size of 20 mm was used in the mix for larger areas of work where the concrete was to be re-cast. For other areas, the size of the aggregate was reduced to match the depth of the breakout. For shallow repairs, a 1:4 mix of cement and graded aggregate up to 5 mm in size was used. The method used was for the edges of the damaged concrete to be cut using a small disc cutter, taking care not to damage the reinforcement. The arrises were detailed to be slightly undercut to improve the mechanical adhesion of the repair patch. The loose and damaged concrete was removed using a breaker, and the steel reinforcement cleaned using a needle gun. A corrosion inhibiting zinc-rich primer was then applied to the reinforcement in some areas and covered with a cement slurry immediately before placing the new ‘high alkalinity’ repair concrete. Finally, the entire structure (original concrete and patches) was coated using a translucent silica paint to unify its appearance.

2.2 Aesthetic Performance

The aesthetic performance of patch repairs applied to historic concrete structures is a key area of concern for the PEPS project as there has been a history of repairs which have had an unacceptable visual impact [11] and damaged their historic aesthetic. The aesthetic performance of patch repairs was assessed on several factors, including how well the repair matches the original concrete in color, texture and profile, as well as how it weathers over time [20]. While there are no specific requirements for color matching concrete and limited data regarding the variability of concrete color, color variations were assessed through calibrated photography and colorimetric analyses using typical descriptions of ΔE relating to the perception of color by a standard observer [26]:

  • 0 < ΔE < 1 observer does not notice the difference;

  • 1 < ΔE < 2 only experienced observer can notice the difference;

  • 2 < ΔE < 3.5 unexperienced observer also notices the difference;

  • 3.5 < ΔE < 5 clear difference in color is noticed;

  • 5 < ΔE observer notices two different colors.

Color calibrated photos of the repairs are shown in Fig. 1 and Fig. 2, and a summary of colorimetric differences between repairs at ENG05 and the adjacent original concrete is shown in Table 1.

Fig. 1.
figure 1

Hand-applied patch repairs G1 (top left), F (top right), P1 (bottom left), and P2 (bottom right).

Full size image
Fig. 2.
figure 2

Form-and-pour repair G2 to corner of wall.

Full size image
Table 1. Summary of colorimetric differences between repairs at ENG05 and the adjacent original concrete.
Full size table

In general, the aesthetic performance of the repairs was poor. During an initial visual assessment, the hand-applied repairs (Fig. 1) were all easily identified due to visible differences in color, texture, and profile. Additionally, the edges of some of the repairs do not appear to be saw-cut and were feather edged, which is a practice that should be avoided [2]. However, repair G2 (Fig. 2) was less easily identified, and this is likely due to the fact that it was a form-and-pour concrete repair and, as a result, had a texture and profile that was much closer to the original concrete. In addition, this repair was in close proximity to vegetation which resulted in a higher degree of biocolonization than was present on the other patch repairs.

Cores taken from the hand-applied repairs (G1, F, P1, and P2) and the form-and-poured (G2) repair all showed a significantly different composition than those taken from the original concrete. Examples of these are shown in Fig. 3, where the difference in the color of the cement matrix between all samples is clearly visible. There are several possible reasons for this discrepancy. Firstly, it is well-known that sourcing aggregates to match those found in historic structures can be challenging due to the closure of quarries and exhaustion of certain supplies. As such, compromises often have to be made to find an alternative which is both technically and aesthetically suitable. Secondly, the shallow nature of the hand-applied repairs meant that no coarse aggregate could be used and the ratio of cement to fine aggregate also had to be adjusted to compensate. However, the form-and-poured repair (G2) also lacks any visible coarse aggregate, despite the fact that its inclusion would be possible due to the larger volume of the repair.

The differences in mix designs means similarity in surface color between the patch repairs and the original concrete is due to the application of mineral silicate paint and/or biocolonization, and not to the selection of similar raw materials. This highlights the importance of having trials and mockups.

Fig. 3.
figure 3

Color corrected images of ϕ50 mm cores from the original concrete (left), P2 (middle), and G2 (right).

Full size image

2.3 Technical Performance

The technical performance of the repairs was mixed. Patches P1 and P2 were found to have significant areas where delamination was suspected during sounding of the repairs with a small hammer, and both had significant cracks that were visible at the interface with the original concrete at the bottom and left-hand edges of the repair (Fig. 1). Patch F had a small area of hollowness in the top right-hand corner of the repair, but no major cracks. Patch G1 was sound with no audible hollowness or visible cracks, though, as with the other hand-applied repairs, the edges were feather edged. Patch G2 was sound on all three faces but had a visible crack at the edge of the repair on one face (Fig. 2) which is possibly the result of shrinkage.

Opening inspections (Fig. 4) and coring of the hand applied repairs (G1, F, P1, P2) revealed that the depths of all of these repairs were relatively shallow and, in some cases, uneven across the patch: G1 5–20 mm, F 5–35 mm, P1 15–20 mm, and P2 15–20 mm. In each case, the depth of repair was only to the surface of the underlying rebar. There was no indication that the reinforcement was undercut to remove the original concrete around the back of the corroded reinforcement, which is recognized in established guidance on concrete repair as being key to ensuring the long-term performance of surface repairs regardless of the degree of corrosion of the reinforcement [8]. The lack of undercutting and full cleaning of the reinforcement was confirmed by consulting as-built documentation photographs, which also show the application of the zinc-rich primer to the front face of the reinforcement only.

However, the thickness of the wall to which three of the five repairs was carried out was only 110 mm, so it is unlikely that it would have been possible to successfully undercut and clean the reinforcements without causing damage so extensive that it would result in demolition and rebuilding of large sections of the wall.

Extensive debonding of the repair material from the original concrete was visible in the opening inspections of P1 and P2, and, to a lesser extent, in patch F also. Pull-off tests [27] were conducted on areas of G1 and F within which no hollowness was detected during sounding, and the results of these are shown in Table 2. To put these results in context, EN 1504–10 [4] specifies a minimum acceptable result of 0.70 MPa for non-structural repairs.

Fig. 4.
figure 4

Opening inspections of repairs G1, F, P1, and P2 (left-to-right).

Full size image
Table 2. Results of pull-off tests carried out in accordance with EN 1542 [27].
Full size table

3 Case Study II: ENG04

3.1 Backgrounds

ENG04 was built between 1928–30 and is located approximately one mile from the sea. The structure is subject to exposure classes XC4, XF1, and XD1 (moderate humidity, exposed to airborne chlorides), and analyses of the concrete at the time of repair reported chloride levels in excess of 0.4% by weight of cement. The depth and quantity of chlorides suggest that, in addition to ingress from airborne chlorides, chlorides were also cast into the concrete during construction - most likely originating from contaminated aggregate (locally sea-dredged shingle). However, the depth of carbonation was relatively low for a structure of this age (approximately 8 mm). Despite this, reinforcement corrosion was present in areas of low cover, resulting in cracking and spalling of the concrete, and trial repairs were undertaken in 2005 to determine if patch repairing would be successful.

The general composition of the concrete was specified as 1:2:4. Shingle aggregate (4-20 mm; dug from around the site) was used with white Portland cement and a brown dredged sand. In 90% of the patches, the steel was too badly corroded and was cut out and new steel inserted. For one of the patches, the thick steel bars were cleaned using high pressure (28 MPa) water jetting, although it is doubtful if this fully cleaned the back of the steel.

Rough sawn boards, chosen to match the board marked impressions in the original parent concrete, were fixed across the repair sites and plugged into the original concrete surfaces, with cotton wool used to seal the joints. The concrete was then poured through a ‘letterbox’ at the top of the formwork and externally vibrated. The boards were removed early from the patches and the surface rubbed down using muslin cloths to remove the outer cementitious layer, thus exposing the aggregate and improving the match to the adjacent weathered concrete. A few of the repairs were mechanically abraded after curing to further remove the cementitious layer and expose the aggregate.

3.2 Aesthetic Performance

The aesthetic performance of the repairs varied as a result of the complications of matching the adjacent original concrete which varied in texture, and the surface finishing techniques. All patches were similar in profile to the adjacent concrete with matching board marks, but had clearly visible edges and, in the case of A9, excess material at the joint between boards of the formwork which did not match that of the adjacent original concrete. All patches had undergone significant weathering (darkening of the surfaces) and had high levels of biocolonization (moss, lichens, algae) on the surface that were similar to that of the adjacent original concrete (Fig. 5).

In order to better assess the underlying surface condition, the patches were manually cleaned with a wire brush (Fig. 6, Fig. 7). Colorimetric measurements were carried out prior to and following cleaning (Table 3). Patch A9 was the closest color match but still had a ΔE suggesting that an unexperienced observer could notice a difference in color between the two materials, patch A6 had a ΔE suggesting a clear difference in color, and patches A5, A7, and A10 had a ΔE suggesting two different colors are observed. However, these color differences are possibly due to inconsistency in the amount of exposed aggregate and cement paste on the surface between the repairs and the adjacent original concrete. Coring at the edge of the repair across the bond line shows a close match in the color of the cement matrix and the type of aggregate used (Fig. 8).

Fig. 5.
figure 5

Form-and-pour patch repairs A5 (top left), A6 (top right), A7 (bottom left), and A9 (bottom right) as found.

Full size image
Fig. 6.
figure 6

Patch repairs A5 (top left), A6 (top right), A7 (bottom left), and A9 (bottom right) following partial cleaning to remove excessive biological growth.

Full size image
Fig. 7.
figure 7

Repair A10 prior to cleaning (left) and after cleaning (right).

Full size image
Table 3. Summary of colorimetric differences between repairs at ENG04 and the adjacent original concrete.
Full size table
Fig. 8.
figure 8

ϕ50 mm core of interface at the bond-line between patch repair A6 and original concrete, photographed from two sides.

Full size image

3.3 Technical Performance

The technical performance of the repairs was generally very good, especially considering the proximity to the sea and the age of the repairs which were assessed after 17 years. Patches A5, A6, A7, and A9 showed no signs of deterioration – there were no cracks, or areas of hollowness or delamination. Coring was only undertaken on three of the patches (A6, A9, A10) and to a depth of approximately 50–60 mm, which was beyond the depth of reinforcement. However, no interface with the original substrate was encountered during coring, and as-built documentation photographs suggests that the depth of the repairs may range from >70 mm to >150 mm. As a result of this and the exposed aggregate finish, pull-off tests were not undertaken at the site. Phenolphthalein applied to the core holes indicated that carbonation of A6 and A9 may have progressed as deep as 30 mm and 20 mm, respectively, which, based on a covermeter survey of the repair area, may be in the region of the original steel reinforcement in some places. However, opening inspections of A6 and A9 revealed the new steel reinforcement was still in very good condition, with no signs of rust staining or cross-section loss.

Patch A10 (in which the original steel reinforcement had been cleaned by water jetting) showed some signs of deterioration, with three significant cracks noted. The first crack was between 0.5–1.0 mm wide, and hollowness was detected while sounding with a small hammer. A hammer and chisel were used to remove the damaged material and investigate the source of the crack. However, phenolphthalein applied to the breakout indicated the concrete was still highly alkaline 15 mm below the surface (Fig. 7), and the crack terminated without any connection to reinforcement. It was concluded that this was not the result of ongoing reinforcement corrosion. A second crack 0.5–1.5 mm on the top-left side was examined and sounded hollow. Removal of the delaminated concrete revealed ongoing corrosion to the underside of an original horizontal reinforcement bar, and phenolphthalein applied to the breakout indicated that the concrete (>35 mm) was fully carbonated. A third crack, 0.5 mm wide, was also noted, but showed no signs of hollowness or delamination, and so was not investigated further.

4 Conclusion

Patch repairs to historic concrete structures at two different sites in England were assessed. This paper details the preliminary investigation of five patches at each of the two sites and presents a comparison of the results and discusses the different repair approaches that were implemented.

At ENG05 an approach of minimal intervention was adopted to preserve as much of the original fabric as possible. However, the practices that were implemented to achieve this are in contrast to established guidance on concrete repair best practice. For example, the decision not to undercut the steel was taken to prevent the loss of historic fabric; the thin wall would have required complete demolition and rebuilding in order to clean the back of the central steel reinforcements. However, removing the original concrete around the back of the corroded reinforcement is recognized as being key to ensuring the long-term performance of surface repairs [8]. Failure to undercut the steel meant that it would also not have been possible to clean the whole circumference of the reinforcement bar. Furthermore, 80% of the repairs were hand applied over exposed reinforcement and, while hand-applied repairs are an accepted practice, they are typically intended for concrete substrates where the steel reinforcement is not visible after preparing the repair area [9]. Unfortunately, the attempt to minimalize the amount of original fabric lost during the initial repair campaign has resulted in repairs with a shorter lifespan, and ongoing corrosion and section loss of the original reinforcement. As such, future repairs will probably be necessary, and this may result in more loss of original fabric.

While ‘like-for-like’ repairs were specified at both sites, examination of cores from the repairs at ENG05 reveal the use of materials which are different from the original concrete, both in appearance and composition. It is unclear why this happened, particularly as aesthetic matching of the repairs was of concern and the choice of materials had been specified.

From a technical standpoint, the repairs to ENG04 have been much more successful, despite being approximately 10 years older and exposed to a more deleterious environment. This is possibly as most of the corroded steel had been cut out and replaced with new clean steel. While the repair process more closely followed established guidance, it did involve the loss of significant amounts of historic fabric in the form of steel reinforcement and contaminated/damaged concrete, both of which had to be replaced with new material. While one of the repairs showed signs of deterioration, the exact cause of this is unclear but this repair was the only one which had used the original (cleaned) steel reinforcements. Further analyses of field data and laboratory samples is ongoing to fully understand this.

Both case studies have highlighted the challenges in achieving a repair which is a successful aesthetic match to original concrete. They demonstrate that Patch repairs will be influenced by the knowledge of the specifier, skill of the site operatives, materials available, design aspects of the original construction, environmental conditions and often other project realities relating to timescale or funding. These need to be considered early in a project to produce on site technically performing patch repairs. Additionally, both case studies have highlighted the issues caused by inadequate cover to reinforcement at the time of construction – common in many early reinforced concrete structures – and the challenges in addressing this during the repair process to provide a repair which has a long lifespan while retaining the original fabric.

Comparing the two case studies highlights the schism which can exist within the field of conservation in respect to the philosophical approach to repair and the extent to which invasive interventions should be undertaken.