Computational Mechanics 26 (2000) 325332 Springer-Verlag 2000

Smeared crack models of RC beams with externally bonded CFRP plates

Abstract Concrete, a complex mix of variously sized aggregates, sand, water, additives and cement binder, is one of the more common engineering materials used for the design and construction of structures and bridges. Concrete is characterised by good compressive strength properties but it demands the use of internal reinforcement, generally in the form of round steel bars, to carry tensile stresses. The strength of the resulting element is dependent on the amount and distribution of steel reinforcement included during construction. It is not however possible to include additional internal reinforcement after construction in the event of the applied loading being increased and therefore consideration must be given to strengthening the structure externally, demolishing it, or conning it to specic usage, for example a maximum weight restriction on a bridge. In circumstances where restricted usage is not practicable structural strengthening is generally more favourable than demolition and replacement. Research in the area of strengthening of existing bridge beams is currently topical in the European Union given recent EU directives, aimed at encouraging free trade and movement of goods and services, which require all bridges to take 40 tonne vehicles. This paper describes the numerical modelling procedures employed, using smeared crack models available in ANSYS V5.4, to capture the load-deformation response and modes of failure, of reinforced concrete beams which have been strengthened, using carbon bre reinforced polymer (CFRP) composite material plates. Experimental verications of these simulations have also been performed and are discussed in the present paper.

Received 12 December 1999

P. Fanning (&) National University of Ireland, Dublin, Ireland O. Kelly Department of Civil Engineering, Earlsfort Terrace, Dublin 2, Ireland The authors would like to record and express their gratitude to Sika Ireland Ltd who supplied the Sika Carbodur System for this project.

1 Introduction Strengthening of beams in exure can be achieved by bonding or anchoring plates, traditionally steel plates, to the tensile face of a reinforced concrete member. The shear strength can be enhanced with plates bonded or anchored to the beam sides in the shear spans. Steel plates, although effective in providing enhanced strength, are difcult to manipulate on site, cannot be delivered to site in lengths greater than about six metres and on exposed concrete surfaces they are particularly prone to corrosion problems. In recent years signicant effort has been dedicated to the use of bre reinforced polymer composite plates in strengthening applications (Meier et al., 1992; Meier, 1996; Emmons et al., 1998). The plate bonding techniques and the strengthened solutions associated with CFRP composite material plates are recognised, in certain circumstances, to be superior to those achieved using steel. In particular the CFRP composite material combines a high specic strength with a low specic density and is available in continuous rolls of almost any length. The greater ease of manipulation and handling results in reduced temporary works requirements, shorter project durations and lower labour costs. Transportation difculties, which generally restrict the length of steel plates to no more than 6 m, and as a result demand that complicated lap joints be fabricated for even medium length beams, are overcome. Both the bre and matrix of the CFRP composite material are inert, consequently unlike steel plate bonded systems corrosion does not pose a threat to the long term durability of the strengthened system. Although this material has recently been used successfully in strengthening schemes in Europe (Midwinter, 1997; Neubauer and Rostary, 1997; Barboni et al., 1997) and worldwide (Nanni, 1995; Sen and Shahawy, 1994) there are many design issues associated with CFRP materials used in plate bonding schemes which remain to be resolved. Hutchinson and Rahimi (1993) and Saadatmanesh and Ehsani (1991) have shown that the CFRP poststrengthening can lead to a catastrophic brittle failure in the form of plate peeling. This is consistent with the work of Quantrill et al. (1995) and Garden et al. (1993). Plate peeling can be overcome by using external anchorage (Spadea et al., 1998), which extend into the compression zone of the beams. Unanchored plate solutions remain attractive as anchorage adds signicantly to both project costs and complexity. Raoof and Zhang (1997) and Zhang et al. (1995) have sought to predict upper and lower bounds for plate peel off loads based on upper and lower

325

326

bound exural crack densities proposed by Watstein and Parsons (1943). Although attractive the range within which plate peel-off is predicted to occur is large and the lower bound solution can be very conservative. Non-linear models for reinforced concrete systems have been evaluated by Kotsovos and Spiliopoulos (1998) and Hofsetter and Mang (1996). Smeared crack models are available in most commercial nite element software. The suitability of these models in capturing the load-deection response and modes of failure of externally strengthened beams was examined by monotonically loading ten instrumented reinforced concrete beams to failure, eight of which were strengthened in exure with different plate congurations, under laboratory conditions and comparing numerical and test data.

2 Laboratory programme: Test beams and strengthening system The mode of failure associated with plate-peel off in shown in Fig. 1. Typically the plate separates from the reinforced concrete beam at its termination point and unzips along its bonded length bringing with it the concrete beneath the tension reinforcement. Three metre long beams, 155 mm wide by 240 mm deep, were selected for the test programme to negate any effects of scale in this sensitive region. A cross section through the beam, Fig. 2, illustrates the internal reinforcement. Three 12 mm diameter steel bars are included in the tension zone with two 12 mm steel bars as compression steel. Ten shear links, formed from 6 mm mild steel bars, are provided at 125 mm centres in the shear spans. Two unplated beams formed a control set, (B1 and B2), and four different plate lengths were used in subsequent pairs of beams, (B3 to B10). Each of the beams was simply supported with a clear span of 2.8 m and loaded symmetrically and monotonically, under displace-

Fig. 2. Reinforcement details for test beams

ment control, in four point bending, with point loads 0.3 m either side of the mid-span location, to failure. The Sika Carbodur System, comprising CFRP composite material plates and a two-part epoxy resin adhesive, provided the external reinforcement on the strengthened beams. The Carbodur S plates, with a longitudinal modulus of elasticity of 155 GPa, a lower bound tensile strength of 2400 MPa with an associated strain to failure of 1.4%, were used in these tests. These plates, consisting of unidirectional carbon bres embedded in an epoxy matrix, were 1.2 mm thick, 120 mm wide and were manufactured by a pultrusion process. Prior to application of the plates the beam surfaces were ground to expose the aggregate and the Carbodur S plates were degreased to provide a clean bond surface. Sikadur-30 adhesive, the thickness of which was controlled by 3 mm diameter granular beads, was applied to the strengthening plates and the ground concrete surfaces and a period of three days was allowed elapse prior to testing. The test programme was designed to assess the strength enhancement offered by anchored plates and also the relative reduction in ultimate load carrying capacity of these beams due to plate peel-off. Beams B3 and B4 had plates bonded along their full length. These plates were thus continuous under the support locations with the support reactions anchoring the plates to prevent plate peel-off. Beams B5 and B6 had 2.03 m long plates centred on the mid-span resulting in 65% of the shear spans being plated, and beams B7 and B8 had 1.876 m plates, (resulting in 58% of the shear spans being plated), similarly arranged. The shortest applied were 1.7 m long, and placed centrally on beams B9 and B10.

Fig. 1. Plate peel-off

3 Experimental test results The mid-span deections for the 3.0 m beams, control and strengthened, are plotted in Fig. 3 with the failure loads and mode of failure documented in Table 1. The unstrengthened beams, B1 and B2, failed at 68.3 and 67.9 kN respectively.

Beams B5 and B6, which were strengthened with 2.03 m long plate lengths, reached loads of 100 and 103 kN respectively. Beams B7 and B8, 1.876 m long plates, failed at loads of 97 and 64 kN respectively. Finally beams B9 and B10 with 1.7 m plates failed at 64 and 82 kN. The mode of failure for each of these beams was due to plate peeling. Failure was sudden and catastrophic in all cases, similar to Fig. 1, and no indication of imminent failure was evident from the load versus deection response. The loss of strength between beams with anchored plates B3, B4 and B5, B6 is of the order of 10% for both beams tested. Beam B7 saw a further albeit small decrease in ultimate capacity, whereas the similarly plated B8 failed Fig. 3. Loaddeection response of test beams before reaching the ultimate load recorded for the control specimens. A further reduction in plate length to 1.7 m shows a signicant loss in beam strength with beam B9 Table 1. Summary of beam test results failing at a lower ultimate load than control beams. The experimental data clearly demonstrates the potenBeam Plate % of shear Ultimate Mode tial for beam strengthening by externally bonded and anconguration span plated load (kN) of failure chored CFRP composite material plates. However, it is also B1 No plate 0.0 68.3 Flexure evident that signicantly reducing the plates lengths can B2 No plate 0.0 67.9 Flexure result in large variations in the ultimate failure loads. It is B3 Anchored plate 100 111 Shear postulated that this is due to the non-homogeneous nature B4 Anchored plate 100 118 Shear B5 2.03 m 65 100 Plate peel-off of reinforced concrete as a material coupled with the plate B6 2.03 m 65 103 Plate peel-off terminating at a point of increased bending moment in the B7 1.876 m 58 97 Plate peel-off shear spans. Nevertheless the consistent nature of the B8 1.876 m 58 64 Plate peel-off failure loads for beam B5 and B6 would suggest that this B9 1.7 m 50 62 Plate peel-off could be avoided with longer plate lengths. B10 1.7 m 50 82 Plate peel-off The failure load and manner of failure is readily determined for an anchored plate but in situations with no plate anchorage alternative methods of calculation are These beams exhibit a large amount of ductility under required. Smeared crack models using the ANSYS nite loading and the mode of failure is characterised by initial element software have been evaluated to this end. yielding of the tensile reinforcement (at approximately 50 to 54 kN) followed by total collapse once the neutral axis 4 of the strengthened beams rises above the compression Numerical modelling techniques steel. A series of three dimensional solid nite element models These results were consistent with classical reinforced of each of the strengthened systems was constructed using concrete beam theory (Kong and Evans, 1987) which the ANSYS nite element software. The four constituent predicted that the yielding of the tensile reinforcement entities of strengthened reinforced concrete beam systems would be initiated at a load of 52 kN and that the capacity comprise the concrete medium, the internal reinforcing in shear was 110 kN. bars, the adhesive and the strengthening plates. Non-linear The ultimate strength of beams B3 and B4 were 118 and material models are required for each of these constituents 111 kN respectively. The additional tensile stiffness pro- except the strengthening plates which can be considered to vided by the strengthening plates keep the neutral axis of behave in a linear elastic manner to failure. the beam lower in the section thereby allowing greater ANSYS (1998) provides a dedicated three-dimensional bending moments to be sustained. Although consistently eight noded solid isoparametric element, Solid65, to model stiffer and stronger, the strengthened beams are less the nonlinear response of brittle materials based on a ductile and the mode of failure was characteristic of a constitutive model for the triaxial behaviour of concrete shear failure in the shear span of the beam. Given that the after Williams and Warnke (1974). The element includes a relatively thin CFRP composite material plate will not smeared crack analogy for cracking in tension zones and a contribute added shear capacity to the section the failure plasticity algorithm to account for the possibility of conload and mode of failure is consistent with design calcu- crete crushing in compression zones. Each element has lations. Therefore provided sufcient anchorage is proeight integration points at which cracking and crushing vided at the plate ends these test results demonstrate that checks are performed. The element behaves in a linear the additional plates provide additional exural capacity elastic manner until either of the specied tensile or such that the limiting strength condition becomes the compressive strengths are exceeded. Cracking or crushing shear capacity of the beam. The beam strength has been of an element is initiated once one of the element principal increased by approximately 70% of its original capacity stresses, at an element integration point, exceeds the tenalthough it is noted that the mode of failure is no longer sile or compressive strength of the concrete. Cracked or ductile. crushed regions, as opposed to discrete cracks, are then

327

328

formed perpendicular to the relevant principal stress direction with stresses being redistributed locally. In the numerical routines the formation of a crack is achieved by the modication of the stressstrain relationships of the element to introduce a plane of weakness in the requisite principal stress direction. The amount of shear transfer across a crack can be varied between full shear transfer (1.0) and no shear transfer (0.0) at a cracked section. The crushing algorithm is akin to a plasticity law in that the once a section has crushed any further application of load in that direction develops increasing strains at constant stress. Subsequent to the formation of an initial crack, stresses tangential to the crack face may cause a second, or third, crack to develop at an integration point. Cylinder splitting and compression tests on concrete core samples taken from the precast beams were undertaken to identify, the tensile strength, the compressive strength, and the elastic modulus of the concrete. In addition the Poisson's ratio and percentage of shear transfer at a crack were required for the numerical model. The internal reinforcement may be modelled as an additional smeared stiffness distributed through an element in a specied orientation or alternatively by using discrete strut or beam elements connected to the solid elements. The beam elements would allow the internal reinforcement to develop shear stresses but as these elements, in ANSYS, are linear no plastic deformation of the reinforcement would be possible. The smeared stiffness and strut modelling options both allow the elasticplastic response of the reinforcement to be included in the simulation at the expense of the shear stiffness of the reinforcing bars. In this case the reinforcement was modelled using strut elements, Link8. These elements are embedded in the mesh of Solid65 elements and the inherent assumption is that there is a perfect bond between the reinforcing bars and the surrounding concrete medium. A linear elastic perfectly plastic material law, described by the elastic modulus, the yield strength and the post-yield stiffness of the material, was used for these elements. The eight noded isoparametric Solid65 element was also used for the 3.0 mm thick adhesive layer. A limiting tensile strength of 45 N/mm2 derived from exural tests on adhesive specimens was specied. Finally four noded isoparametric linear elastic shell elements, Shell63, with orthotropic material properties accounted for the strengthening plates.

Fig. 4. Solid65, link8, and shell63 elements of the nite element mesh

Two planes of symmetry necessitated only a quarter model and a typical nite element mesh, with a proportion of the concrete elements removed to reveal the internal reinforcement, the adhesive and the strengthening plates, is shown in Fig. 4. The material properties for all model constituents are summarised in Table 2.

5 Correlation of experimental & numerical results The load versus deection response at midspan, for the numerical models of the control beams, B1 and B2, and the strengthened beams with anchored plates, B3 and B4, are plotted with the experimental data in Fig. 5. The numerical model of the control beam indicates a failure load of 55.8 kN compared to 68 kN obtained in the laboratory. Yielding of the tensile reinforcement occurs at a load of 51.6 kN in the numerical model which is consistent with the location of change of slope in the load deection response of the test beams. The numerical crack distributions and experimental crack distributions near midspan, at failure, are given in Fig. 6. The numerical model predicts that exural cracks, overlaid on the nite element mesh and illustrated using circles parallel to the crack plane, extend uniformly through the depth of the beam (hence at 90 to the plane of the page) before becoming less uniform as the compression

Table 2. Material properties

Material properties Elastic moduli (Gpa) Poisson's ratio Tensile strength (Mpa) Compressive strength (Mpa) Shear transfer coefcient Yield strength (Mpa) Post-yield stiffness (Gpa)

Concrete 39.2 0.27 5 80 1.0 NA NA

High yield steel 210 0.3 NA NA NA 460 0

Mild steel 210 0.3 NA NA NA 250 0

Adhesive 11.6 0.3 45 NS 1.0 NA NA

CFRP composite material plates 155 (Long.) 10 (Trans.) 0.3 (Major) 0.0194 (Minor) NA NA NA NA NA

NA: not applicable; NS: not specied

Fig. 5. Numerical and experimental deection comparisons B1, B2, B3, B4

face is approached. This is also evident in the test beam. However the discrete nature of the exural cracks is not captured using a smeared crack model. The mode of failure predicted using the numerical model was a exural mode of failure, consistent with the test response, due to increasing plastic strains developed in the tension reinforcement. For the strengthened beams with anchored plates the numerical model became unstable at 106.5 kN. At this load exural cracks had extended through the depth of the beam and the compression strength of the concrete material had almost been reached in the compression zone. The concrete elements which have cracked, in at least two directions, at each integration points are highlighted in Fig. 7a for comparison with the test failure mode in Fig. 7b. The damaged area is approximately 330 mm from the load point. The mode of failure in the test beams was a catastrophic failure originating at locations 160 and 235 mm, for beams B3 and B4 respectively, to the left of the roller load point in Fig. 7b. Subsequent to this crack

forming the section lost stability and the plate was pulled from under the support point thereby forming the large distinct crack evident in the tension zone of the beam in Fig. 7b. The numerical ultimate load of 106.5 kN compares favourably with the test results of 111 kN and 118 kN, for B3 and B4 respectively, and the locations of origin of the damaged areas are also consistent. The correlation between test and numerical strains for measurement points at the centre of the plate and on the top surface of the concrete, both at midspan, is shown in Fig. 8. The correlation for both plate and concrete strains are good although the fall in gradient of the plate strains in the numerical models, as a result of the formation of cracks in the adhesive, is not evident in the test results. However in general the numerical models are simulating quite closely the response of both the unstrengthened beams (B1, B2) and the strengthened beams with anchored plates (B3, B4). The test and numerical deection responses for beams B5, B6 are plotted in Fig. 9. The correlation between test and numerical data is again good. The ultimate numerical load was 103.5 kN, compared to values of 100 and 103 kN reached in the tests. Although the concrete between the bonded plate and the tensile reinforcement was cracked in three principal directions it was not evident from the simulation that failure was due to plate peel-off. Reducing the plate lengths in the numerical models to capture the response of beams B7, B8 and B9, B10 resulted in ultimate failure loads of approximately 96 kN in both cases. The numerical models did however capture the deection and plate strains, Fig. 10, in the constant bending moment region up to the load at which plate peeloff occurred in the test beams. The formulation of the material model includes a parameter to model the extent of shear transfer across an open or closed crack. Attempts

329

Fig. 6. B1 and B2 experimental & numerical crack distributions at failure

330

Fig. 7. a Location of initial damage in FE model (B3, B4); b beams B3, B4 localisation of failure

Fig. 8. Strains test plate (B3-P), concrete (B3-C) strains and FE data

Fig. 9. Numerical and experimental deection comparisons B5, B6

to capture the reduced failure loads by varying this parameter between full shear transfer and 10% of full shear transfer were not successful. The strains of 50 mm from the plate ends are plotted in Fig. 11. The correlation of plate strains at this location is again seen to be good, up to the point at which the plate peels off, although the strains predicted by the numerical model are slightly higher. Furthermore experimental plate strains for beams B9 and B10 are also consistent up to the point at which rapid crack propagation occurs initially. The deviation in ultimate loads for these similarly plated

beams is attributed to different rates of crack propagation at the plate ends due to the non-homogenous nature of the surrounding concrete material. The smeared crack model tackles the formation of a crack by assessing the principal stress values and directions at element integration points. If a principal stress is greater than the tensile strength of concrete the element stiffness is reformulated so as to have zero stiffness in that principal stress direction. Stiffness in the remaining principal stress directions are maintained and any stresses

Fig. 10. Comparison of numerical and experimental midspan plate strains, B9, B10

The effectiveness of the external plates in strengthening the reinforced concrete beams was found to reduce as plate lengths were shortened. In the case of strengthened systems having plates bonded over 65% of their shear spans, strength enhancements of 50% over the strength of the original beams were achieved. However when the percentage of the shear span which was plated was reduced to 50% the strength enhancement, for one beam of a pair, was 22% with the second beam failing at a lower load than the unstrengthened system. Anchored plates thus represent the optimum strengthened solution and strengthening schemes using unanchored plates, should only be implemented where it is possible to extend the bonded plate signicantly into the shear spans. Smeared crack models for reinforced concrete beams have been demonstrated to capture the progressive ductile failure mechanism of unstrengthened beams. The distribution of cracked elements in the strengthened beams with anchored plates was similarly found to be consistent with the shear mode of failure in the test beams. The random and more dynamic nature of crack propagation associated with plate peel-off modes of failure are however nor captured using a smeared crack model and it is recommended that these modelling techniques be restricted to schemes which include anchored plates.

331

Fig. 11. Plate strains 50 mm from plate curtailment, beams B9, B10

References

in the cracked principal stress direction are redistributed and an equilibrium solution is sought. Cracks are thus formed in a uniform and controlled manner in the numerical model and no account is taken of dynamic crack propagation. Additionally the non-homogeneous nature of concrete as a material is not represented and as a result equilibrium solutions beyond the point of plate peel off in the experimental tests are possible.

6 Conclusions The test programme clearly demonstrates that the potential for CFRP composite materials in strengthening schemes is signicant. The strength of the 3.0 m reinforced concrete beams, in four point bending, were increased by up to 70% of their original value while the stiffness was increased by approximately 40%. Examination of the crack distributions under loading indicated that the size and distribution of cracks in the strengthened beams were also less signicant than those recorded at equivalent load points in the control beams. It is also evident from the test data that the mode of failure and the load at which failure occurs is dependent on the length of plate and whether or not the plate is anchored. The largest strength enhancement is achieved with anchored plates and the resulting full composite action between the existing beam and the strengthening plates allows classical reinforced concrete design calculations to be extended to reliably predict the ultimate load and mode of failure.

ANSYS, ANSYS Manual Set, 1998, ANSYS Inc., Southpoint, 275 Technology Drive, Canonsburg, PA 15317, USA Barboni M, Benedetti A, Nanni A (1997) Carbon FRP strengthening of doubly curved precast PC shell. J. Composites for Construction 1(4): 168174 Emmons PH, Vaysburd AM, Thomas J (1998) Strengthening Concrete Structures, Part 2, Concrete International, April, pp. 5660 Garden HN, Hollaway LC, Thorne AM (1993) A Preliminary Evaluation of Carbon Fibre Reinforced Polymer Plates for Strengthening Reinforced Concrete Members. Proc. Instn. Civ. Engrs. Structs. & Bldgs. vol. 123, pp. 127142 Hoffsetter G, Mang HA (1996) Computational plasticity of reinforced and prestressed concrete structures. J. Comput. Mech. 17(4): 242254 Hutchinson AR, Rahimi H (1993) Behaviour of Reinforced Concrete Beams with Externally Bonded Fibre Reinforced Plastic. Proc. 5th Int. Conf. on Structural Fault and Repair, pp. 221228, Edinburgh, 1993 Kong FK, Evans RH (1987) Reinforced and Prestressed Concrete, 3rd Edition, Van Nostrand Reinhold (UK) Kotsovos MD, Spiliopoulos KV (1998) Evaluation of structuralconcrete design concepts based on nite element analysis. J. Comput. Mech. 21(4/5): 330338 Meier U (1996) Advanced Composite for Structural Repair: European Perspective. Proc. US European Workshop on Bridge Engineering, pp. 491504 Meier U, Deuring M, Meier H, Schwegler G (1992) Strengthening of Structures with CFRP Laminates: Research and Applications in Switzerland. Proc. Ist Int. Conf. on Advanced Composite Materials in Bridges and Structures, Sherbrooke, Canada pp. 243251 Midwinter KR (1997) Plate bonding carbon bre and steel plates. Construction Repair Nanni A (1995) Concrete Repair with Externally Bonded FRP Reinforcement: Examples from Japan, Concrete International, pp. 2226

332

Neubauer U, Rostasy FS (1997) Strengthening of Bridges with Externally Bonded CFRP-plates. Proc. 7th Int. Conf. on Structural Faults and Repairs, Vol(2), pp. 211220, Edinburgh Quantrill RJ, Hollaway L, Thorne AM, Parke GAR (1995) Preliminary Research on Strengthening of Reinforced Concrete Beams using GFRP, Non-Metallic (FRP) Reinforcement for Concrete Structures. Proc. 2nd Int. RILEM Symp (FRPRCS2), Gent Raoof M, Zhang S (1997) An Insight into the Structural behaviour of Reinforced Concrete Beams with Externally Bonded Plates. Proc. Inst. Civ. Engrs. Structs. & Bldgs. Vol. 122: pp. 477492 Ritchie PA, Thomas DA, Lu LW, Connelly GM (1991) External reinforcement of concrete beams using bre reinforced plastics. ACI Struct. J. 88(4): 490500 Saadatmanesh H, Ehsani MR (1991) RC beams strengthening with GFRP plates 1: experimental study. J. Struct. Eng. ASCE 117(11): 34173433 Sen R, Shahawy M (1994) FRP Research for Highway Applications in Florida, Proc. 3rd Materials Engineering Conference,

Infrastructure: New Materials and Method of Repair, pp. 280 286, San Diego Spadea G, Bencardino F, Swamy RN (1998) Structural behaviour of composite RC Beams with externally bonded CFRP. J. Comp. Construction 2(3) Watstein B, Parsons DE (1943) Width and spacing of tensile cracks in reinforced concrete cylinders. J. Res. Natl Bur. Standards 31(RP545) 124 William KJ, Warnke ED (1975) Constitutive Model for the Triaxial Behaviour of Concrete, Proceedings of the International Association for Bridge and Structural Engineering Vol 19, p. 174, ISMES, Bergamo, Italy Zhang S, Raoof M (1995) Prediction of Peeling Failure of Reinforced Concrete Beams with Externally Bonded Steel Plates. Proc. Inst. Civ. Engrs. Structs. & Bldgs. Vol. 110: pp. 257268 Zhang S, Raoof M, Wood LA (1997) Discussion: Prediction of Peeling Failure of Reinforced Concrete Beams with Externally Bonded Steel Plates. Proc. Inst. Civ. Engrs. Structs. & Bldgs. 122, pp. 493496