Crystallized alkali-silica gel in concrete from the late 1890s

https://doi.org/10.1016/j.cemconres.2006.05.017Get rights and content

Abstract

The Elon Farnsworth Battery, a concrete structure completed in 1898, is in an advanced state of disrepair. To investigate the potential for rehabilitation, cores were extracted from the battery. Petrographic examination revealed abundant deposits of alkali silica reaction products in cracks associated with the quartz rich metasedimentary coarse aggregate. The products of the alkali silica reaction are variable in composition and morphology, including both amorphous and crystalline phases. The crystalline alkali silica reaction products are characterized by quantitative X-ray energy dispersive spectrometry (EDX) and X-ray diffraction (XRD). The broad extent of the reactivity is likely due to elevated alkali levels in the cements used.

Introduction

The Elon Farnsworth Battery has overlooked the entrance to Portsmouth Harbor for over 100 years. Artillery has been part of the landscape since 1631, when the first guns were installed in an earthen redoubt [1], [2]. Placement of the concrete for Farnsworth Battery began in 1897, at a time when the use of Portland cement was gaining in popularity over the more commonly used natural cements [natural cement, or “rock cement” was typically produced from a single source of limestone that contained enough impurities to create a product similar to modern Portland cement] [7]. The history of the construction of the battery was preserved in the lettercopy book of Inspector J.W. Walker. Excerpts from his notes, compiled by Perrault and Ofenstein in a historical report on the battery, reveal that two natural cements from the New York Rosendale District were used during construction, Hoffman Cement and Beach's Cement, as well as one Portland cement from the Pennsylvania Lehigh Valley, Atlas Cement [3]. The natural cements were primarily used in the bulk concrete of the parapet. The Portland cement was reserved for the remaining concrete construction [3]. Table 1 lists chemical compositions for the three cements, as reported in literature from the period [4], [5]. The most striking compositional difference between the cements is the high MgO content of the natural cements versus the Portland cement, (Table 1). Also of interest in Table 1 is the high level of alkalies reported for the Hoffman Cement. Natural cement from the Rosendale District was widely used in the 1800s and early 1900s. By the 1920s, Portland cement dominated cement production, and the use of natural cement declined [6]. In the late 1890s, there were at least fifteen different companies producing natural cement in the Rosendale District, but the practices of the Lawrence Cement Company, producer of the Hoffman brand, are well documented and briefly described here [4], [8]. The rock used in Hoffman cement was quarried from a blend of an upper, light colored limestone, and a lower, dark colored limestone, both separated by a layer of sandstone. In the morning, alternating layers of crushed rock and anthracite pea coal were fed into the top of a 3 m dia. by 8.5 m high kiln, and burnt overnight. The following morning, the burnt product was discharged from the base of the kiln and sorted. Over-burnt rock was discarded, and under-burnt rock was returned to the top of the kiln. Properly calcined rock was crushed, ground, and packaged in barrels. The practices of cement production in the Rosendale District remained relatively unchanged since its inception in the 1830s. Atlas Cement of the 1890s, on the other hand, represented the latest in cement technology, and was the first successful Portland cement produced in the United States by rotary kiln [9], [10]. The coarse aggregate used in the battery, a quartz rich metasedimentary rock, originated from crushed material excavated from the site. The fine aggregate, according to the inspector's notes, was shipped from Plum Island, Massachusetts, a source 30 km to the south [3]. Also according to the inspector's notes, water for the concrete mix, as well as for moist curing, was taken directly from the harbor [3]. After completion of the battery, it was noticed that the guns in their retracted position protruded as much as 20 cm above the parapet, so, in 1903, an additional 23 cm of concrete was added to the parapet apron [3]. Fig. 1 shows the battery as armed in 1898. The battery experienced structural and moisture problems soon after construction, and the guns were removed during World War I. In 1948, the fort was deactivated [11]. In 2001, a 2 ha parcel of land, including the battery, was transferred to the University of New Hampshire (UNH). To assess the condition of the battery, cores were collected from the concrete in 2003. Fig. 2 shows the current condition of the battery.

It is widely accepted that the production of alkali silica gel plays the primary role in the deleterious nature of the alkali silica reaction. Alkali silica gels may crystallize over time, and observations of crystalline alkali silica reaction products have been reported throughout the literature [12], [13], [14], [15], [16], [17], [18], [19], [20], [21], [22], [23]. Most researchers attribute the presence of crystalline alkali silica reaction products to the crystallization of alkali silica gel. A variety of mechanisms have been reported to explain the crystallization. Some researchers have found that crystallization requires the drying of the gel. Cole and Lancucki reported the crystallization of gel after heating to 110 °C, followed by a period of 12 days at ambient room conditions [13]. Fernandes et al. describe a gel that crystallizes over time when kept in a desiccation chamber [22]. Some researchers have reported compositional differences between gels and crystallized gels [15], [16], [17], [18], [22]. Thordal Anderson and Thaulow, and Thaulow et al. noted that crystallized gels are often observed in cracks within the aggregate, and less often observed in cracks within the cement paste [17], [18]. They suggest that compositional differences between the crystallized gel and the amorphous gel may be due to local differences in the pore water chemistry within the aggregate versus the pore water chemistry within the cement paste. Kurtis et al. describe an experiment where the crystallization of alkali silica gel is observed in situ via soft X-ray transmission microscopy when immersed in solutions of calcium and sodium hydroxide [19], [20]. It is the experience of the authors that the quantities of crystallized gel observed in concrete are generally limited as compared to normal, or amorphous alkali silica gel. However, crystallized gel is abundant in the concrete of Farnsworth Battery, and is present in quantities suitable for collection and study. The characterization of the crystallized gel is the primary focus of this paper.

Section snippets

Experimental

Four cores were received in an air dry state: two from the shell room beneath gun position number 1, one from the parapet apron in front of gun position number 1, and one from the floor adjacent to gun position number 1. An additional half core, split during unconfined compressive strength testing by UNH, from the parapet wall adjacent to gun position number 1 was also received. Macro examination of the cores showed abundant cracks and voids filled with white alkali silica reaction product. The

Petrographic microscope observations

Fig. 3, Fig. 4, Fig. 5 show deposits of alkali silica reaction products as observed in thin section. In Fig. 3, a large deposit of crystallized gel is situated between two coarse aggregate particles. In Fig. 4, a crystallized gel deposit sits in a crack within a coarse aggregate particle. Fig. 5 shows a large crack through a coarse aggregate particle. The crack exits the aggregate, and extends into a large gel deposit within a void in the cement paste. In Fig. 5, crystallized gel lines the

Discussion and conclusions

The chemical composition of the crystallized alkali silica gel from Battery Farnsworth is similar to the chemical composition of crystallized gel reported by Cole and Lancucki, which they identified as okenite, with K and Na substitutions [13]. However, there are some questions whether the okenite structure could accommodate such extensive substitution of monovalent K+1 and Na+1 for divalent Ca+2. The mineral formula of the crystallized gel from the battery, based on 32 anions, was computed to

Acknowledgements

Thank you to Pete Payette of the American Forts Network, and Bolling Smith of the Coast Defense Study Group, Inc., for permission to use current and historic images of the battery. Thank you to Fred Bailey, who recorded the diffraction pattern used in this paper.

References (25)

  • M.S.J. Gani

    Cement and Concrete

    (1997)
  • L.C. Sabin

    Cement and Concrete

    (1905)
  • Cited by (53)

    View all citing articles on Scopus
    View full text