In situ concrete sewer performance: comparison of Portland cement, calcium sulfoaluminate cement, and calcium aluminate cement

This paper provides a better understanding of the performance of calcium sulfoaluminate cement (CSA) in comparison with calcium aluminate cement (CAC) and Portland-limestone cement, CEM II A/L 52.5 N (CEM II) in live sewer environments. Three concrete mixes using these binders, with 0.34 w/b, siliceous pit sand, and dolomite aggregates, were prepared for two years of exposure in two sewer sites. During exposure, monitoring through visual observation, concrete surface pH, mass and thickness change was conducted regularly to observe the deterioration. At the end of exposure, microstructural analysis, i.e., Scanning Electron Microscopy, Quantitative Evaluation of Minerals by Scanning Electron Microscopy (QEMSCAN) and XRD analyses, were conducted to elucidate the deterioration mechanisms further. Regardless of sewer exposure conditions, the monitoring results indicated that CAC concrete had superior performance, followed by CSA, then CEM II. However, corrosion rates were accelerated when a sewer had hydraulic actions and high H 2 S gas concentrations (max.>300 ppm). CSA outperformed CEM II due to the presence of ettringite, while ettringite superposition in the transition zone reduced CSA performance compared to CAC concrete. Binder performances were primarily related to their chemistry, mineralogy, and interaction with aggregates.


Introduction
Under certain sewer conditions, concrete pipe systems are susceptible to severe deterioration, resulting in structural and functional failure within their service life.This deterioration is exhibited in the sewer pipes and manholes, and wastewater treatment plants [1].According to Kaempfer and Berndt [1], 40% of the deterioration is attributed to Biogenic Acid Corrosion (BAC), while 60% is ascribed to inadequate structural design, construction, installation, and control and maintenance.If engineering experts are well-informed, 60% of the failure can be eliminated.However, the challenge remains to minimise the 40%.
Biogenic acid Corrosion (BAC) -also known as Microbially Induced Corrosion (MIC)results from biological activities of Sulphide Oxidising Bacteria (SOB) on the unsubmerged concrete surfaces, which produces acid that corrodes the concrete microstructure and the underlying steel reinforcement.A sewer pipe contains biodegradable organic matter with sulphate compounds in a slime layer between the submerged sewer concrete surface and wastewater, the so-called 'biofilm'.In favourable conditions of temperature and insufficient dissolved oxygen, anaerobic reactions in the biofilm occur in the presence of Sulphate Reducing Bacteria (SRB) to reduce oxygen from the sulphate compounds.Consequently, hydrogen sulphide (H2S), a very corrosive gas, is produced [2,3].
Turbulent flow conditions and high velocities of some sewers facilitate the release of this gas into the sewer headspace (atmospheric sewer space above the wastewater).H2S is oxidised to sulphuric acid (H2SO4) by SOB (abiotic and biotic processes).Subsequently, the acid reacts with calcium silicate, calcium aluminate, and carbonate compounds of concrete to form gypsum and ettringite.Both gypsum and ettringite are expansive products leading to internal cracking and spalling of concrete, which provide sites for further acid penetration, concrete pH reduction and SOB proliferation.Thus, concrete slowly loses its structural integrity and exposes the underlying steel reinforcement to corrosion [3][4][5].
Cracks, erosion, structural spalling, and corrosion of concrete and reinforcing steel can lead to wastewater leakage, eventually resulting in air, water, and soil pollution.Kaempfer and Berndt [1] estimated that, in Germany alone, about 500 million m 3 /year of contaminated waters with sulphate, chloride and nitrogen compounds are associated with sewer leakage, posing a threat to the environment and human health.In addition, sewer deterioration substantially increases the direct and indirect costs attributed to sewer system repair and rehabilitation, leading to productivity losses of other functioning entities and lost wages due to social service interruptions [6][7][8].
Various repair and rehabilitation techniques have been employed to combat corrosion.For instance, lining techniques, mainly thermoplastic-based liners, are widely applied to avoid the disruptions associated with excavating and replacing sewer pipes and increasing sewer longevity.However, they are deemed expensive and less effective in certain situations, such as concrete surface profiles and infiltration between the liner and host pipe [9,10].Also, they might cause massive aggressive gas accumulations in sewers leading to corrosion of unprotected elements such as manholes and air pollution in case of any opening.
Other preventive measures, such as the application of chemical or biological technologies (that decrease H2S production and emission) and protective coatings and acid-resistant cement (that prevent the chemical attack of concrete), have been investigated [11][12][13][14].However, most protective materials, such as epoxy coatings and antimicrobial polymer fibres or metal-zeolites, are claimed to perform well in the laboratory but with little or no evidence of their field performance.Available field performance indicates that epoxy coatings are notorious for delaminating and peeling from the concrete substrate, antimicrobial agents tend to dissolve in wastewater [14], and many cementitious coatings may perform poorly [11,13].
These inter-related problems can be overcome by developing or improving cement-based lining systems or producing corrosion-resistant concrete pipes manufactured using available highperformance binder systems, such as Alkali-Activated Cements (AAC), Calcium Aluminate Cement (CAC), Calcium Sulpho-Aluminate cement (CSA), and belite-Y'elimite cement.These binders are expected to perform better because their chemistry and microstructure should impart more excellent acid resistance.However, implementing these binder systems requires understanding their deterioration mechanisms and potential to resist and possibly suppress BAC in a live sewer environment.
Therefore, this paper aims at a better understanding of the performance of CSA and compares its performance to a known superior sewer-performing binder (CAC) and the less wellperforming Portland cement binder (CEM II) when subjected to two live sewer environments [15].

Materials
This study used three binders: CEM II/A-L 52   Three concrete mixes with 0.34 w/b, shown in Table 2, were designed using the C&CI method [16] and cast following a typical vibration compaction technique.Except for the binders, the constituent contents were kept constant for all mixes, intended to reduce performanceinfluencing variables.Superplasticiser (BASF Master Glenium 7700 ) with a 0.3 -0.5% dosage by cement volume was used during mixing to enhance concrete compaction.Concrete specimens for Durability Index (DI) tests [17] and field exposure were cast in cylindrical moulds, 100 mm in diameter by 300 mm in height.For saturated density and compressive strength tests, 100 mm cubes were used.After compaction, the specimens were left to set and harden for 24 hrs, after which they were immediately demoulded and heat-cured for another 48 hrs.The heat-curing technique involved specimens wrapped in air-and watertight plastic bags and immersed in hot water in a curing tank maintained at 40±2°C for 48 hrs.Subsequently, the specimens were air-cured in a room maintained at 20±3°C and 50% relative humidity (RH) to achieve similar maturity characteristics as sewer concrete pipes.

Laboratory quality control tests
Laboratory quality control tests (compressive strength, saturated density, and DI), were conducted before exposing the specimens to the sewer sites.The compressive strength and saturated density were determined after 28 days of curing, according to SANS 5863 [18].
DI specimens were prepared as per SANS 3001-CO3-1 [19].The tests, i.e., the Oxygen Permeability Index (OPI) test and Water Sorptivity Index (WSI) test, were performed per SANS 3001-CO3-2 [20] and the UCT DI manual [17], respectively 2.4 Specimen preparation for exposure and monitoring Specimens for exposure were prepared and exposed for monitoring, as indicated in Bakera et al. [21].The monitoring included visual observation, mass and dimension measurements, and surface pH measurements, which were conducted monthly to quarterly for about two years.

Sewer site background and environmental conditions
Two sewer sites in the City of Cape Town in South Africa were considered for concrete exposure, i.e., Northern Area Sewer Manhole 19 (NAS) and Langa sewer pump station manhole (LPS).Detailed information regarding NAS is provided by Bakera et al. [21], and those for LPS in Kiliswa [3].LPS manhole for exposure is situated between two pump stations; because of the periodic pumping effect, specimens experience wetting and drying cycles, i.e., sewer hydraulic actions.
Gas concentrations of both sites changed diurnally, seasonally, and sometimes due to recurrent sewer occurrences.The frequent pumping effect causes the variation of the gas at the LPS, while at NAS, it is caused by the peak hours of wastewater flow, with the highest peaks in the morning and evening.The maximum gas concentrations were about 360 ppm and 105 ppm for LPS and NAS, respectively.Both sites indicated lower gas concentrations during the cold seasons and higher concentrations in the hot seasons.and XRD analysis, were carried out at the end of the exposure cycles to assess the effect of BAC on the concrete microstructure.After gently removing organic debris on the deteriorated surfaces, the specimens were submerged in an isopropanol solution for seven days to displace free water, followed by three days of oven-drying at 40 o C, and then split in half; one half for SEM-BSE and QEMSCAN analysis and another for XRD analysis.As in Bakera et.[21], SEM-BSE and XRD analyses were conducted, while QEMSCAN analysis was as in Bakera and Alexander [22].

Results
The following results were obtained after exposing concretes to two sewer sites, NAS and LPS, for two years.

Saturated densities and compressive strengths
After 28 days of curing, concrete average saturated densities were 2574 ± 5 kg/m 3 , 2575 ± 3 kg/m 3 , and 2529 ± 22 kg/m 3 for CEM II, CSA, and CAC, respectively.These densities are above 2500 kg/m 3 and similar to those obtained in previous studies [3,23] for concrete with similar aggregates.
Compressive strength results were 59.2 ± 1.5 kN, 70.2 ± 0.8 kN, and 89.0 ± 2.6 kN for CEM II, CSA, and CAC, respectively.These strengths were within the expected range of concrete sewer pipes, i.e., 50 MPa to 80 MPa [23], except CAC, with the highest compressive strength, which was attributed to the conversion of metastable hydrates (C2AH8 and C4AH13) to stable hydrates (C3AH6 and AH3 ) at a curing temperature of 40 o C [24].
According to Alexander et al. [17] and Moore et al. [26], CEM II is considered excellent quality concrete, with a WSI value of 5.5± 0.1 mm/hr 0.5 and 6.4 ± 0.4% porosity.CSA and CAC are regarded as excellent to good quality concrete because of their WSI values of 8.9 ± 0.4 mm/hr 0.5 and 7.3 ± 0.5 mm/hr 0.5 , and 3.5 ± 0.2% and 8.5 ± 0.4% porosity, respectively.CSA had the lowest porosity because of ettringite formation, which has a high expansion volume.Generally, all concrete was of good quality for sewer exposure.

Visual observations
Visual observation results are shown up to 26 months of exposure at NAS but only to 15 months at LPS because LPS specimens were unavoidably lost on site before the full exposure period.This hazard can exist when working in 'real' sewer environments where unexpected operational conditions can sometimes arise.

At NAS
Fig. 1 shows visual observations of concrete at the NAS.Before exposure, the surfaces of CEM II were predominantly grey, and those of CSA and CAC were light yellow, with a clear distinction between cement matrix and aggregate particles.
These specimens were initially exposed in the summer period.After eight months, CEM II experienced corrosion on the cement matrix, with aggregate particles, i.e., dolomite, showing minor corrosion.The cement matrix changed to a whitish colour, i.e., corrosion deposits, mainly observed on the edges of the specimens.By contrast, CSA and CAC showed black stains, indicating microbial colonisation, with no sign of corrosion deposits on the exposed surface.Since the concrete specimens were exposed during mainly colder seasons up to 8 months, it can be postulated that the aggressivity of the site was mild because of low H2S gas concentrations (see Section Error!Reference source not found.).
At 15 months of exposure, all concretes, except CAC, had blisters on the epoxy-coated curved surfaces.CEM II exhibited severe signs of corrosion, including aggregate surface depressions and disintegration and significant whitish paste-like deposits on the exposed surface.CSA and CAC had no sign of corrosion besides dark stains on their surface, probably biofilm, and some whitish spots on their edges.Generally, CEM II was more severely corroded after this duration.
After 26 months, the entire concrete matrix was severely corroded with significant material loss.Yellowish stains demarcating severely corroded surfaces and whitish deposits were observed on CEM II [27].Dolomite aggregates slightly protruded on the exposed surface, as also observed by Hudon et al. [28], showing signs of dissolution.The aggregate protrusion indicates that its deterioration rate was slower than the cement matrixes.Significant material losses were observed on CSA specimens, starting from the edges to the centre.At the same time, its cement matrix changed to brownish and then whitish as a sign of corrosion deposits.CAC concrete indicated mud-like deposits on the exposed surface, which changed into whitish corrosion deposits towards the edges.The epoxy coating on the CAC specimens remained intact, and no significant material loss was observed.Based on this site, it was concluded that, visually, CAC had superior performance, followed by CSA, and then CEM II.

At LPS
Fig. 2 shows visual observations of concrete at LPS for up to 15 months.After eight months of exposure, CEM II surfaces changed to yellowish with black stains on the edges and aggregate protrusion.CSA and CAC showed minor deterioration and black deposits on the surface, with brownish stains on CAC only.Due to the effect of sewer hydraulic actions at this site, no corrosion products were observed on the surface of all concretes.
At 15 months, all concretes experienced signs of material loss, aggregate protrusion and fallout, and blisters on the epoxy surface.The rates of cement matrix dissolution contrasted from one binder system to another, with CEM II severely corroded.CAC concrete showed superior performance to other concretes.Its cement matrix slightly deteriorated on the exposed surface, illustrated by the remaining epoxy coating after corrosion product removal.This implies that CAC concrete does not entirely resist corrosion but has the slowest rate of deterioration.Similar visual observations were reported by Khan et al. [29].CSA deterioration rate (i.e., material removal on the exposed) was somewhat higher than CAC but better than CEM II.Generally, it can be concluded that CSA and CAC perform better under sewer hydraulic actions than CEM II.
The LPS site was more aggressive than the NAS due to its sewer hydraulic actions.Consequently, significant corrosion product removal, renewal of exposed surfaces, and aggregate protrusion and fallout were observed on concrete surfaces.

Concrete surface pH
Microbial proliferation in sewer conditions controls concrete surface pH and deterioration [29,30].Therefore, BAC stages induced by sewer aggressivity and microbiological activity [31] can be studied using concrete surface pH.Fig. 3 shows concrete surface pH measurements of specimens exposed at the NAS for 26 months and at the LPS for 15 months.
At the commencement of exposure, all concretes had surface pH between 10.0 and 8.8, lower than fresh concrete, possibly due to carbonation during sample preparation.CSA had the lowest surface pH, followed by CAC and CEM II.
After eight months, all concretes had a surface pH of around 6, except CEM II at NAS. Neutrophilic SOBs were presumed to have colonised the concrete surface, priming it for acidophilic SOB colonisation [30].CEM II at NAS had a higher pH than the other concretes due to its higher CaO content, which increases neutralisation capacity.After 15 months, all concretes had pHs between 6 and 4, indicating increased sewer aggressivity and corrosion deposits on the exposed surface, which shields the "fresh" substrate from pH measurement.At NAS, CEM II had pHs between 5 and 4.5, and CSA and CAC had around 4. At this stage, CAC had the lowest pH, possibly linked to decreased calcium ion leaching [32] and gibbsite production, which is stable at pH 3-4.LPS concrete, in contrast, had surface pHs between 5.5 and 5.0 higher than NAS concrete because of corrosion deposit removal due to sewer hydraulic actions [33] This implies that concrete surface pH does not necessarily reflect sewer aggressivity but rather the interaction between concrete alkaline reactive components and acid production.
At 26 months, NAS concrete pH values decreased sharply, notably for CEM II, which had pHs below 3. CSA and CAC pHs were 4-3.At these pHs, acidophilic SOBs had completely colonised the surfaces.CSA and CAC pH values agree with the literature, attributed to gibbsite formation on the exposed surface [24].As explained earlier, no readings were taken at LPS at this age because the specimens were lost.
Therefore, it can be concluded that the gradual evolution of concrete surface pH is observed with time due to successive microbial colonisation, leading to aggressive acid generation and continuous corrosion deposition in the absence of severe sewer hydraulic actions.Concrete surface pH also implies interaction between concrete reactive components and acid production, that is, SOB produce acid at a certain low pH, but after interacting with the concrete reactive components, the measured pH represents a neutralised pH.

Mass changes
Fig. 4 demonstrates the mass changes of concrete specimens exposed at NAS for 26 months and at LPS for 15 months.

At NAS
At NAS, CEM II showed significant mass loss within eight months of exposure, while CAC and CSA showed a slight increase in mass gain up to 15 months, similar to the observations of Khan et al. [29] on CAC mortar after 12 months of sewer exposure.The mass gain is ascribed to moisture absorption, adhering organic matter, and sulphate penetration into the concrete microstructure [29].Moisture absorption and sulphate penetration on a thin outer surface layer (transition zone) may lead to anhydrous phase hydration and secondary ettringite formation, hence further mass gain.Ettringite formation may induce microcracks in the microstructure, accelerating degradation.However, after long-term exposure, the initial mass gain may have little effect, and can likely be disregarded.All concretes experienced mass loss after 26 months, with CAC showing the lowest mass loss, followed by CSA, and CEM II the greatest, which correlated with visual observations.Corrosion on the exposed surfaces led to cement matrix and aggregate loss.In the case of CSA, ettringite decomposed to gypsum and gibbsite, which are low strength, hence the mass loss.CAC's high porosity, which could accommodate ettringite expansive pressure, and its chemical behaviour are possible reasons for its reduced mass loss.Generally, CAC portrayed the best performance in terms of mass loss, and CEM II had the worst performance.

At LPS
At LPS, all concrete experienced significant mass loss by eight months of exposure, contrasting with the CAC and CSA results at NAS.Such significant mass losses were chiefly associated with high H2S gas concentration and the sewer hydraulic action at this site.As corrosion occurred, sewer hydraulic actions continuously removed corrosion products and protruding aggregate, renewing the exposed surface [34,35].Significant aggregate protrusion and fall-out occur when the cement matrix deteriorates faster than the aggregates, hence the higher mass loss.A uniform corrosion front, with less aggregate protrusion and fallout, was observed when both cement and aggregate matrix had a similar deterioration rate, hence lower mass loss.This scenario was visually evident in the CSA and CAC concrete, where less aggregate protrusion was observed.
Comparing concrete performance at the NAS and the LPS sites, the combined corrosion and sewer hydraulic actions leading to significant aggregate fallout at the LPS site accelerated the rate of deterioration.
Generally, also from the visual observations, CAC concrete showed superior performance, followed by CSA, then CEM II at both sites.CSA's better performance compared with CEM II in terms of mass loss corresponds with the results obtained by Yang et al. [36] and Cao et al. [37] under sulphuric acid attack.

Thickness change and corrosion rate
Thickness losses after the respective exposure durations and their equivalent corrosion rate are summarised in Table 3 (For more details see Online Resource 1).At NAS, all concretes experienced thickness loss, which accelerated after about eight months of exposure.However, CAC concrete showed minor loss throughout the exposure duration, with thickness losses from the lowest to the highest CAC <CSA < CEM II.At LPS, all concretes exhibited continuous thickness loss after eight months of exposure, with a similar rank as at NAS. Comparing mass change with thickness change results, CEM II had significant corrosion deposits on the exposed layer, counteracting the mass loss at NAS, whereas, at LPS, this was washed away.Despite CSA concrete exhibiting no mass loss up to 15 months of exposure, it experienced thickness loss, implying that the mass gain due to the formation of secondary ettringite or corrosion products compensated for the mass loss due to sewer corrosion and loss of thickness.
Presenting the influence of binders in resisting BAC in percentage terms, the performance of CSA was 50% higher, and CAC was 94 ± 2% higher than CEM II at NAS, and 54 ± 8% and 71 ± 6% at LPS, respectively.Based on this, it can be concluded that CAC has the best performance, and CEM II has the worst.The performance of the CSA binder ranks between that of CAC and CEM II.

Microstructural analysis at NAS
The microstructural analysis results, i.e., SEM-BSE images, QEMSCAN analysis, and XRD analysis, are discussed for the specimens exposed to NAS for 26 months (i.e., about two years) because those at LPS were unavoidably lost on site before the end of exposure duration.

CEM II
A typical SEM-BSE image for CEM II after 26 months of exposure to NAS is shown in Fig. 5. Fig. 5 (A) shows the intact and altered zones.The intact zone is dense, and the altered zone is porous, with fragmented fine particles detaching from the intact cement matrix.The altered zone depth is about 2.1 mm in the cement matrix and about 1.1 mm in the aggregate matrix, indicating a non-uniform corrosion front.On the other hand, coarse aggregate (dolomite) disintegrates in three layers (Fig. 5 (B).The first layer at the exposed surface is weak and porous, followed by a disintegrated layer with superior integrity and cohesion.The final layer is separated by a crack from the intact particle, with its surface dissolved and disintegrated.Magnitude-wise, the concrete microstructure mainly consists of about 80% aggregate components (dolomite and quartz).In the cement matrix, the percentage weights of Quartz particles, on the other hand, are associated with a silica-rich layer [28] following Ca-, Al-and Mg-silicate disintegration in the deteriorated zone Legend Fig. 6 QEMSCAN analysis of CEM II concrete subjected to the NAS for two years.The legend indicates the colour code, percentage weight and percentage volume of each phase over the crosssection.Other images depict the distribution of selected phases on the cross-section XRD analysis (Online Resource 2) confirms that the intact zone consists of portlandite and calcium silicate in the cement matrix and dolomite and quartz phases from the aggregates.After acid penetration, the exposed surface and 100 µm depth from the exposed surface indicate significant gypsum peaks, which decrease at 200 µm depth.Some Mg silicate peaks are also observed in the altered layer, as in the QEMSCAN analysis.
Therefore, the main cement phases for CEM II before acid attack are calcium silicate and portlandite, which contain some aluminium, iron, and magnesium oxides.Once sulphuric acid is generated on the exposed concrete surface, it decalcifies calcium silicate and portlandite [27,38].As a result, calcium ions react with sulphate to form gypsum, magnesium ions migrate into the transition zone to react with silicate, then iron and aluminium ions in the deteriorated zone form Fe/Al-silicate, while some ions form gibbsite.Thus, gypsum is observed in the transition zone along with Mg-silicate, with the deteriorated zone exhibiting gibbsite and

CSA
Fig. 7 shows the SEM-BSE image for CSA concrete after 26 months of exposure at NAS.The intact zone consists of a very dense microstructure with some voids and significant coarse aggregate content.There is no clear demarcation between the intact and altered zone.The estimated altered zone depth is about 0.9 mm, almost uniform in the cement and aggregate matrices, thus indicating similar aggregate and cement deterioration rates.Fig. 7(B) shows continuous microcracking parallel to the exposed surface, yet not fully demarcating the deteriorated zone from the transition zone., ettringite formation, due to CSA hydration and sulphate penetration, as pointed out by Aboulela [39].CSA concrete microstructure also has a high concentration of aggregate phases, followed by ettringite, Ca-aluminate, gibbsite, Portlandite, and gypsum.
Gibbsite is a part of hydration products when ye'elimite only hydrates in the presence of gypsum [40][41][42].After sulphate penetration, ettringite decomposed to produce more gibbsite and gypsum in the deteriorated zone since ettringite is unstable in a low pH environment [40,41].
Intact zone, 2.0 mm Altered zone, avg.0.9 mm The XRD analyses (Online Resource 3) show the presence of ettringite, gibbsite, calcium aluminates, and other phases, such as anhydrite, ye'elimite, and calcium iron oxide (CaFe2O4), in the intact zone, as reported in the literature [36,[43][44][45].The altered zone comprises gypsum, aluminium oxide, magnesium-bearing phases, dolomite, and quartz peaks.Gypsum formation indicates the dissolution of the calcium-rich phases.In contrast, magnesium phases, such as magnesium iron silicate, magnesium aluminium silicate, and Hercynite-magnesia, possibly imply interaction between the CSA cement and dolomite constituents, yielding these phases.It, therefore, suggests carrying out further study to investigate this hypothesis.
Under XRD analysis, gibbsite was not readily observed in the deteriorated zone, although significantly mapped under QEMSCAN analysis.According to Aboulela [39], only amorphous gibbsite is formed in the deteriorated zone of CSA concrete, which could not be detected under XRD analysis.This is because amorphous phases produce wide X-ray scattering profiles, which are difficult to detect in XRD patterns with high peaks of crystalline phases [46].

CAC
The microstructure of CAC concrete is shown in Fig. 9, with a dense microstructure and some voids in the intact zone, see Fig. 9 (A).The altered zone consists of a few fragmented cement particles with a maximum depth of 1 mm, less than that observed in the other concretes.A uniform and even corrosion front is observed, implying that the cement and aggregate matrix have equal deterioration rates.Organic matter is observed at the interface between the altered and intact zone; see Fig. 9(B).Clearly, identifying the transition zone under this analysis was also challenging.QEMSCAN analysis in Fig. 10 indicates that CAC concrete microstructure also consists of quartz and dolomite aggregates in high concentrations, followed by, in order, Ca-aluminate, ettringite, gypsum, gibbsite, portlandite, and Ca-silicate.The main phases in the intact zone are Ca-aluminate, Portlandite, gibbsite and Ca-silicate, supported by XRD results in Error!Reference source not found..In the deteriorated zone, a significant amount of gibbsite and gypsum is observed, as well as some minor phases such as Fe/Al silicate and Mg-silicate.
As pointed out earlier, it was challenging to distinguish between the deteriorated and transition zones in the SEM-BSE image.However, in QEMSCAN analysis, a distinct layer of secondary ettringite was observed in the transition zone, which was associated with sulphate penetration into the concrete matrix due to the gypsum reaction with calcium aluminate hydrate [47].
Nevertheless, no ettringite was observed within the deteriorated zone because it occurs only in the deeper sections of the concrete microstructure, with a pH higher than [28,48,49].Also, Kiliswa [3] indicated that CAC concrete tends to have a broader and more cohesive altered zone than PC-based concrete.The broader zone may be attributed to ettringite formation.However, this observation contradicts the literature that no ettringite is formed in the CAC concrete [50].Additionally, it is possible that this broad zone is more tenacious due to its high pH and hence contributes to the better performance of CAC concrete.Still, these points need to be clarified with further analysis and study.Under XRD analysis (Online Resource 4), the intact concrete consisted of gibbsite, Katoite, and calcium aluminium oxide, which all fall under the Ca-aluminate mapped in the QEMSCAN analysis.At the exposed surface, the main phases observed were gypsum, quartz, and dolomite, while ettringite and gibbsite peaks were not observed.At 200 µm and 500 µm depth from the exposed surface, minor peaks of ettringite were observed, though this was inconclusive since prominent peaks were not observed.As it was difficult to analyse the microstructure below 0.5 mm, proving the presence of ettringite in the transition zone was impossible.However, the XRD analysis results by Khan et el.[29] indicated ettringite presence within a 3 mm depth of CAC mortar specimens after being subjected to a sewer environment.Therefore, further study is recommended to confirm this.

Influence of sewer exposure
BAC in a live sewer environment is generally a slow process [29,[51][52][53], which requires a long-term study of several years to determine a steady and roughly constant average corrosion rate [54].Therefore, the average corrosion rates presented in Concretes at LPS exhibited higher corrosion rates and mass loss than those at the NAS because of higher H2S gas concentration and sewer hydraulic actions (wetting and drying cycles, as well as hydraulic erosion).
Sewer hydraulic actions at LPS influenced continuous corrosion layer removal and protruding aggregate fallout, renewing the exposed surface, consequently accelerating the deterioration.Aggregate fallout, on the other hand, depends on the cement deterioration rate.For binders with higher deterioration rates than the aggregates, significant aggregate fallout is experienced, and vice versa.
Concretes at NAS experienced mass loss due to BAC only, which resulted in substantial microbial activities for acid production and subsequent corrosion product build-up.The corrosion products were occasionally removed on site due to gravity and occasional sewer flooding and, in the laboratory, during the cleaning of specimens for bulk measurements.
These observations indicate that the performance of binders is highly dependent on the sewer aggressivity, which includes the H2S gas concentration, temperature, and sewer hydraulic action, provided that there is sufficient relative humidity and oxygen supply for oxidation.

Deterioration mechanisms
Based on the concrete microstructure analyses, BAC affects different concretes differently, depending on the binder chemical composition, hydration phases, and cement-aggregate interaction.The SEM-BSE and QEMSCAN analyses indicate that in addition to the concrete depth 'lost' on-site over the exposure period, the remaining concrete consists of two sub-zones in the altered zone, i.e., the transition zone and the deteriorated zone.These zones define the extent to which acid has penetrated the concrete microstructure.The deteriorated zone is usually less cohesive, while the transition zone retains a measure of integrity, cohesiveness and lower porosity, but less than the intact zone.Because of this, the transition zone is difficult to observe in SEM analysis.However, with QEMSCAN analysis, it can be identified with sulphur-bearing phases such as gypsum and ettringite, and in some binders with Mg-silicate and gibbsite layer, indicating calcium-deficient layers [15] The following discussion, therefore, addresses phase evolution in the concretes containing CEM II, CSA and CAC under BAC.

CEM II binder system
The main hydration products of CEM II binder systems are Portlandite and hydrated calcium silicate, both of which are susceptible to acid corrosion.They react with sulphuric acid to produce gypsum and ettringite, which are expansive and have lower mechanical properties.As a result, the concrete becomes more porous and degrades quickly [37,55], thus, explaining the poor performance of CEM II in this study.

CAC binder system
CAC is well known as a superior acid-resistant binder due to the long chain of dissolution of its hydrates, leading to more gibbsite formation, high neutralisation capacity and gibbsite stability at pH between 3.5 and 4. The literature also indicates that the corrosion end products, mainly gibbsite and stratlingite, offer some mechanical or physical integrity to the altered zone, which provides a barrier to further corrosion [15,29].
In this study, QEMSCAN analysis indicated that the main hydration products of CAC concrete were calcium aluminate hydrate (CA and C3AH6) and gibbsite, which, with acid penetration, form a broad layer of secondary ettringite in the transition zone, followed by gibbsite and gypsum in the deteriorated zone.However, ettringite was not confirmed under XRD analysis as the transition zone was not analysed due to 'smothering' by the high aggregate content.
The ettringite layer in the transition zone was presumed to be the layer observed by Kiliswa [3] and Khan et al. [29].This raises concerns about the layer's stability since secondary ettringite, which is expansive, can induce microcracking in the concrete microstructure, thus accelerating deterioration.It is thus hypothesised that a significant amount of gibbsite in this layer (that provides physical integrity) retains some layer integrity, hence offering microstructure endurance and resistance.Fig. 11 illustrates a chain of hydration, dissolution and evolution of CAC phases under BAC, established based on the CAC hydration reaction in Scrivener et al. [56].

CSA Binder system
The main hydration products of CSA concrete are ettringite, gibbsite and calcium aluminate hydrates, and some residual portlandite, gypsum and Ye'elimite.Under BAC, ettringite decomposes to form gibbsite and gypsum in the deteriorated layer since it is unstable at a pH lower than 10.7.However, with acid penetration, secondary ettringite is formed in the transition zone from the gypsum and calcium aluminate hydrate reaction [47].As a result, the transition zone in the concrete microstructure is complex to define due to ettringite superposition, i.e., ettringite originally present before deterioration and ettringite formed subsequently after sulphate penetration [21].
CSA microstructure also has low porosity due to the original ettringite volumetric expansion.Secondary ettringite formation promotes microcracking within the microstructure, amplifying the attack.In that case, ettringite superposition may explain why CSA has poorer performance than CAC concrete, despite the presence of gibbsite.Fig. 12 indicates a hypothesis of CSA hydrates' deterioration mechanism under BAC.

Influence of aggregates on BAC
When acid-reactive aggregates are incorporated into concrete, they effectively retard the progress of concrete deterioration because they corrode along with the cement matrices, leading to a slower corrosion rate and, occasionally, a more uniform corrosion front [57].A uniform corrosion front occurs when the deterioration rates of the cement and aggregate matrices are approximately equal or match.An uneven corrosion front is observed when these deterioration rates are mismatched, leading to aggregate protrusion or fallout and an accelerated concrete corrosion rate.Therefore, aggregate and cement matrix deterioration rates that are comparable or similar should considered when selecting aggregates for best sewer concrete performance.
In this study, CAC and CSA had reasonably uniform corrosion fronts, suggesting that their cement and dolomite matrices had roughly equal deterioration rates.In contrast, CEM II exhibits uneven corrosion fronts, with its altered zones having higher depths in the cement matrixes than the dolomite aggregate particles.This observation suggests that the CEM II deterioration rate is higher than that of dolomite aggregate, which is usually observed in practice.Therefore, the superior performance of CAC and CSA likely to result from good compatibility with dolomite aggregates.
Additionally, dolomite is a combination of calcium carbonate and magnesium carbonate.When it reacts with the acid, gypsum is formed, while its magnesium carbonate dissociates into magnesian ions since magnesium sulphate (MgSO4) does not precipitate due to its high solubility [58].Therefore, it is possible that the Mg-silicate observed in the transition zone of CEM II originated from dolomite.

Conclusions
Based on the findings presented in this paper, the following can be concluded.
i) Due to corrosion deposits, organic debris, and acid penetration, sewer concrete may experience some initial mass gain, which would have minimal effect on the long-term performance after several years of exposure.
ii) Regardless of the sewer environmental conditions described in this paper, CAC concrete exhibits superior performance, followed by CSA and CEM II; this observation is consistent across techniques such as visual observation and mass and thickness change.With sewer hydraulic actions and high H2S gas concentration, the concrete corrosion rate is accelerated.
iii) The susceptibility of calcium silicate hydrates and portlandite to BAC leads to CEM II overall performing relatively poorly.iv) CSA outperforms CEM II due to the presence of ettringite, which has a higher neutralisation capacity than calcium silicate and portlandite, and its transition to gibbsite, which is stable at pH 3-4.However, ettringite superposition in the transition zone, due to its expansive nature, which induces microcracking, causes lower CSA performance than CAC concrete.v) CAC's better performance is due to 'the long chain of dissolution of its hydrates, which increases gibbsite production and neutralisation capacity.CAC also tends to have a broader transition zone filled with ettringite due to its stability at high pH, which with acid penetration, yields more gibbsite in the deteriorated zone.
vi) This study detected magnesium silicate in the CEM II transition zones, possibly originating from dolomite breakdown after acid penetration.Further study on cement matrix-dolomite aggregate interactions under BAC is recommended.
Finally, these results indicate that the three binder systems exhibit different performances in different live sewer environments, primarily related to the chemistry and mineralogy of the binder, coupled with its interaction with aggregates.Therefore, the selection of binders should be related to sewer aggressiveness, with high-resistance binders used in more aggressive sewer environments.

Fig. 2
Fig. 2 Visual observations of UCT concrete specimens exposed at the LPS manhole for 15 months

Fig. 3
Fig. 3 Concrete surface pH readings of specimens exposed at the NAS and LPS.The error bars represent the standard deviations of the pH measurements.

Fig. 4
Fig. 4 Mass change of concrete specimens exposed at the NAS and LPS.Positive (+ve) values indicate mass gain, and negative (-ve) values indicate mass loss.The error bars represent the standard deviations of the mass change measurements.

Fig. 5 Fig. 6 ,
Fig.5SEM-BSE image for CEM II concrete after two years of exposure at NAS. B is the zoomed section in A Fig.6, the QEMSCAN analysis of CEM II, defined the altered zone as a combination of the transition and deteriorated zones.The intact zone mainly contains calcium silicate and portlandite in the cement matrix.The transition zone consists of a coherent layer of gypsum and Mg-silicate deposits, and the deteriorated zone has finely distributed quartz particles.Some particles are detached from the transition zone and cracked.

Fig. 7
Fig. 7 SEM-BSE image for CSA concrete after two years of exposure at NAS: B is the zoomed section in A QEMSCAN analysis of CSA concrete (Fig 8) also does not clearly demarcate the deteriorated and transition zones.This is possibly associated with the superposition of ettringite in the transition zone, i.e., ettringite formation, due to CSA hydration and sulphate penetration, as pointed out by Aboulela[39].CSA concrete microstructure also has a high concentration of aggregate phases, followed by ettringite, Ca-aluminate, gibbsite, Portlandite, and gypsum.

Fig 8
Fig 8 QEMSCAN analysis of CSA concrete subjected to the NAS for 26 months The legend indicates the phase's colour code, percentage weight and volume over the cross-section.Other images depict the distribution of selected phases on the cross-section

Fig. 9
Fig. 9 SEM-BSE image for CAC concrete after two years of exposure at NAS. B is the zoomed section in A

Fig. 10
Fig. 10 QEMSCAN analysis of CAC concrete subjected to the NAS for 26 months.The legend indicates each phase's colour code, percentage weight and volume over the cross-section.Other images depict the distribution of selected phases on the cross-section.

Fig. 11
Fig. 11 Hydration, dissolution, and deterioration mechanism hypothesis of CAC under BAC Note; ettringite is formed only in the transition zone.

Fig. 12
Fig. 12 Hydration, dissolution, and deterioration mechanism hypothesis of CSA under BAC.Note CSA hydration takes place in three ways depending on the amount of CH and CS̅ .

Resource 1 Online Resource 2
Thickness of the sample specimens exposed at the NAS for 26 months and LPS for 15 months.Positive (+ve) values indicate thickness gain, and negative (-ve) values indicate thickness loss -XRD analysis of CEM II concrete subjected to the NAS for two years, at the exposed surface, at 100 µm and 200 µm from the exposed surface, and the intact concrete Online Resource 3 XRD analysis results of the CSA concrete subjected to the NAS for two years.atthe exposed surface, 100 µm and 200 µm from the exposed surface, and the intact concrete Online Resource 4 XRD analysis results of the CAC at the exposed surface, 100 µm and 200 µm from the exposed surface, and the intact concrete

Table 1
Chemical composition of various binders and fine and coarse aggregatesAggregates were used; siliceous pit sand (PS) from the Malmesbury-Klipheuwel sand deposit in Cape Town and dolomite aggregate (DA) (coarse and fine) from the Olifantsfontein quarry in Gauteng Province, South Africa.The chemical compositions of binders and aggregates are presented in

Table 2
Concrete mix design with 0.34 w/b

Table 3
Thickness losses and equivalent average corrosion rates of concrete specimens subjected to the NAS and the LPS site for 26 months and 15 months, respectively.While the figures above are given to 2 decimals, this should not be construed as a measure of accuracy for the long-term since more long-term results would have been preferable.

Table 3
only indicate the initial corrosion stage, which may change before approaching a constant value in long-term exposure.