Engineering Structures 40 (2012) 479486

Contents lists available at SciVerse ScienceDirect

Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct

The Maria Pia Bridge: A major work of structural art q

A.P. Thrall a,, D.P. Billington b, K.L. Bra c
a

Department of Civil and Environmental Engineering, Princeton University, Engineering Quadrangle E321, Princeton, NJ 08544, USA Department of Civil and Environmental Engineering, Princeton University, Engineering Quadrangle E323, Princeton, NJ 08544, USA c Skanska USA Building, Empire State Building, 350 5th Avenue, 32nd Floor, New York, NY 10118, USA
b

a r t i c l e

i n f o

a b s t r a c t
This paper presents a technical, historical, and aesthetic study of the Maria Pia Bridge over the Douro River in Porto, Portugal which was designed and built by G. Eiffel and Cie. between 1875 and 1877. Through these analyses, this paper demonstrates the signicance of this bridge due to its (1) economy as shown through the design competition, (2) efciency and safety under self-weight, live, and wind loads as revealed by nite element analyses, and (3) elegance of form which is evaluated through the aesthetic motivation of the designer, international acclaim, and an on-site visual analysis. The primary purpose of the paper is not to suggest that designers today emulate the Maria Pia form, but to shed light on the thought process of the engineers at G. Eiffel and Cie. The conceptual design was imagined through considering the site constraints, the forces acting on the structure, and the erection procedure, ultimately leading to an economical project that won the design competition. This design-build approach can be used more often in 21st century design. 2012 Elsevier Ltd. All rights reserved.

Article history: Received 12 August 2011 Revised 6 February 2012 Accepted 7 February 2012 Available online 10 April 2012 Keywords: Arches Bridges Competition Historic sites Iron

1. Introduction Gustave Eiffel (18321923) and his company, G. Eiffel and Cie., designed and built the Maria Pia Bridge over the Douro River in Porto, Portugal (completed in 1877) (Fig. 1). It is a two-hinge, iron arch bridge with a span of 160 m and a rise of 42.935 m [1]. It was the rst of Eiffels great arch bridges. The second, the Garabit Viaduct (completed in 1884), is better known but follows closely its design and construction. This paper will demonstrate the signicance of the bridge through its economy, efciency, and elegance. A detailed description of the design competition will highlight the economy of the design. Technical analyses reveal the structure to be efcient and safe under self-weight, live, and lateral wind loadings. The elegance of the design is demonstrated through the aesthetic motivation of the designer, international acclaim, and an on-site visual study. Finally, a discussion of the relevance of this research to 21st century bridge design will be presented. The main purpose of all of these studies will be to demonstrate the design and construction process of the primary engineers: Gustave Eiffel and Thophile Seyrig (18431923). They formed G. Eiffel and Cie. in 1868 a partnership which lasted for 8 years and
q

resulted in the design and construction of many works, including the Maria Pia Bridge [2,3]. Seyrig wrote a report on the analysis of the bridge titled Le Pont sur le Douro Porto which this paper relied on for details related to the structure [1]. This paper will present the analysis techniques and the design-build mentality of both engineers which were fundamental to winning the design competition and to the success of the nal built structure. 2. The design competition The economy of the design is shown through the design competition held by the Royal Portuguese Railroad Company that opened on May 1, 1875. Four companies (Mead, Wrightson and Co., Socit des Batignolles, Cie. Fives-Lille, and G. Eiffel and Cie.) collectively provided six designs. Seyrig lists each design anonymously, but Loyrette (1985) and Cruz and Cordeiro (1981) provided details related to the companies and costs of the projects [2,4]. Eiffels proposal (costing 965,000 francs) was 31% less in cost than the next lowest design [1]. The nal cost exceeded the original bid, but was still 9% less than the next lowest design [4]. See Fig. 2 for elevation drawings of each proposal and the designer and cost (when known). The site for the bridge crossing a 400 m wide, 61 m deep valley between Via Nova de Gaia and Porto would complete a rail link between Lisbon and Porto, but posed an engineering challenge due to soil instability at the riverbed. Any alternative site would add an additional 12 km of travel [1]. The 150 m wide Douro River, that runs through this valley, ranges in depth from 15 to 20 m with

Corresponding author. Present address: Department of Civil Engineering and

Geological Sciences, University of Notre Dame, 159 Fitzpatrick Hall, Notre Dame, IN 46556, USA. Tel.: +1 5746312533. E-mail addresses: [email protected] (A.P. Thrall), [email protected] (D.P. Billington), [email protected] (K.L. Bra). 0141-0296/$ - see front matter 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.engstruct.2012.02.032

This document is a collaborative effort.

480

A.P. Thrall et al. / Engineering Structures 40 (2012) 479486

Fig. 1. Maria Pia Bridge (foreground). Photograph by Bra.

(A)

Mead, Wrightson and Co. 2,750,000 F

(B) (C)

Socit des Batignolles 1,895,000 F

Socit des Batignolles

(D)

Cie. Fives-Lille 1,410,000 F

spring from these piles to the masonry abutments on either side of the valley. The arches were to be erected simultaneously starting at each side of the river such that the half-arches would balance the central arch during erection [1]. Socit des Batignolles proposed two designs featuring bowstring trusses. The rst, less expensive design spanned the entire river with a 170 m bowstring truss with a 22.5 m depth at midspan (Fig. 2B). The main truss members of this design would be in steel rather than iron. Seyrig raised concern over the design with respect to potential dangerous oscillations from wind excitation. Furthermore, the center of the gravity of the bowstring truss is signicantly below deck level. With wind loads around 270 t, the resulting moment from the wind could result in the uplift of the entire structure. The bowstring truss is supported by masonry piers which might not be able to sustain the high lateral wind loads [1]. Socit des Batignolles second design (Fig. 2C) featured a 120 m bowstring truss with a depth at midspan of 15 m. This design resulted in an in-river pier requiring a foundation in the sandy riverbed [1]. This second design was highly problematic both with respect to its higher cost in comparison to their rst proposal and due to the difculty with the in-river pier foundation. Cie. Fives-Lilles proposed the second least expensive proposal (Fig. 2D) consisting of two metal piers on either side of the river with two 86 m struts supporting the central span. The deck is then divided into four equal spans of 78 m. The large surface area of the deck compared to the other designs would result in high wind forces in the structure. The same company submitted a second, more expensive project (Fig. 2E) which places a pier in the river. This design is subject to the same problems with foundation design as previously discussed [1]. There were two other designs originally proposed prior to the opening of the competition, but were withdrawn (Fig. 2F and G). See Seyrig (1878) for a review of these designs [1]. G. Eiffel and Cie.s design was chosen from among these proposals on the basis of cost.

(E)

Cie. Fives-Lille

3. Erection of the bridge One of the advantages of G. Eiffel and Cie.s design was its erection method. This design crossed the river by means of a central arch. The 354.38 m long deck is supported at ten points, including the crown of the arch and the abutments. The bridge was constructed from 1450 t of iron, 750 t of which lies in the arch alone. To avoid the unstable soil conditions, no supporting piers or falsework were placed in the river-bed [5]. The erection sequence of the arch began with the construction of the piers on either side of the river which were built with the use of cranes (Fig. 3A). Arch panels were pre-assembled and then lifted into place, where they were suspended by cables and pivoted from the main supports (Fig. 3B). Eight cables on either side of the river supported the cantilevered arch during erection. The arch panels were mounted in succession until they met at midspan when the nal keystone was added (Fig. 3C). When the two halves came together there was a negligible horizontal deviation at the center [5]. Eiffel stated that, At the junction of the two intrados half-arches at the keystone, we could observe that the work in the shop and the assembly had been done so carefully that, between the two arch portions, the horizontal deviation was limited to approximately one centimeter, and it disappeared as a result of the assembly pins [5]. This attests to the precision of G. Eiffel and Cie.s workmanship and design. The deck was assembled on embankment platforms and was launched as the arch construction progressed towards the center [5]. The deck was launched with the use of rollers and powered only by human labor at a rate of 9 m/h [4].

(F)

(G)
Fig. 2. Proposals for design competition including designer and cost (in Francs). G. Eiffel and Cie.s proposal cost 965,000F [2,4]. Image reprinted from Seyrig [1].

high river currents and is subject to ooding that can elevate the water level by 10 m [5]. Nearly 15 m beneath the water surface, surveys had revealed the existence of a thick layer of sand. At the time, it was less expensive to divide a span longer than 150 m into two smaller spans. However, these soil conditions would make in-water piers inadvisable. Though bridge spans had been increasing steadily, none had reached 160 m [1]. This competition provided the impetus for engineers to consider this challenge. Mead, Wrightson and Co. proposed the most expensive design (Fig. 2A). This design featured a 160 m central arch resting on masonry piers. On either side of the arch, two 82.5 m half-arches

A.P. Thrall et al. / Engineering Structures 40 (2012) 479486

481

Fig. 3. Erection sequence. Image reprinted from Eiffel [5].

4. Detailed design of the arch and original calculations The span and the depth of the arch were dictated by the site constraints. The shape of the arch itself is derived from preliminary calculations on the bridge. Initially, a parabolic arch was assumed, but high bending moments resulted. Instead, the shape of the arch was eventually selected through hand sketches which balanced a reasonable distribution of bending moments with the designers interest in developing an elegant shape. With the shape of the arch dened, a crescent form (meaning an arch that is narrowest at the hinge and deepest at the crown) was selected since the highest bending moments would be experienced to the left and right of the crown. The greatest depth of the arch was therefore assigned to the crown and the crescent form followed from there. The crescent form had previously been employed for roof beams in train stations, but this is the rst time that it was used for an arch bridge. Out of the plane of the bridge, the arch is widest at the base and narrows toward the crown to counteract the effects of wind and horizontal oscillations from train trafc [1]. Even with these characteristics of the arch pre-determined, as Seyrig points out in his article, there is no direct method to determine the cross-section properties on the rst attempt. The selfweight of the arch is of course initially unknown as is the distribution of the material in the arch. A trial and error approach, as noted by Seyrig, would be long and tiring. Instead, a combination of methods was employed to arrive at a detailed design of the arch truss [1]. First, Bresses theory of circular arches was employed to determine a preliminary value of the arch thrust [1,6]. To employ this theory, it is assumed that the neutral ber of the arch is close to that of a circular arch, that the cross-section is symmetric about the neutral axis, and that there is not much variation of the moment of inertia along the arch (although the arch deepens at the crown, the cross-sectional area increases toward the hinges making the the moment of inertia sufciently close to constant) [1]. With these assumptions, Seyrig asserts that it is then possible to employ Bresses tables for the design of circular, iron arches with constant cross-section [1,6]. Using these tables, Seyrig nds a preliminary value for the thrust and therefore all external loads acting on the bridge (including self-weight from the deck) are known. Then the arch is divided into two systems: (1) one arch with vertical and diagonal members and (2) the same form with vertical and diagonal members inverted. Graphic statics is then employed to

calculate the stress in each member for each of the two systems. The two systems are then superimposed to nd the total stress in each member [1]. Fig. 4 shows drawings highlighting the use of graphic statics in the design of the Maria Pia Bridge. Due to the prominence of the structure and the its long span, however, a special commission was formed to investigate its safety and stability. This investigation employed the force method of analysis with numerical integration to make the nal detailed calculations on the arch. More specically, this analysis method was employed to calculate the horizontal reactions and stresses in the arch under various loadings, including self-weight, three static live loads, and wind loads. For example, to calculate the horizontal reaction, the horizontal degree of the freedom at the hinge was released and a given load was applied to the arch. The horizontal translation can be calculated by considering the effects of bending and axial compression on each section of the arch. Then the load is removed and a horizontal thrust is applied. The horizontal translation due to only the thrust can be found in the same manner. By requiring that the total horizontal displacement be zero, the horizontal thrust can be determined. For such calculations, each half of the arch was divided into 11 sections and the section properties of each segment were projected onto the neutral axis. See Table 1 for the geometry and section properties of the arch. Numerical integration was then employed for the calculations [1]. See Winter et al. (1964) for a review of this method of analysis [7]. See Thrall (2008) for a more detailed discussion of how this method applies to Seyrigs calculations [8]. Note that both axial and bending deformations were considered in this method. If the axial effects had been neglected, the horizontal thrust under selfweight would have been just 1.5% different. This difference in horizontal thrust is effectively negligible and shows that axial effects could have been ignored altogether. Alternatively, as Parker discusses in his 1881 thesis on the bridge, the horizontal reactions

Table 1 Geometric properties of arch. The rst column indicates the node (see Fig. 5), the second and third columns indicate the horizontal and vertical coordinates, respectively, the fourth column provides the cross-sectional area, the fth column provides the moment of inertia in the plane of the arch, and the sixth column provides the moment of inertia perpendicular to the plane of the arch. Data reprinted from Seyrig (1878) Tables 1 and 14 [1]. Node 1 2 3 4 5 6 7 8 9 10 11 x (m) 2.80 8.40 14.10 20.40 25.25 31.00 39.75 49.15 59.20 69.60 80.00 y (m) 3.00 9.00 14.55 20.42 24.20 28.30 32.75 36.85 40.35 42.25 42.65 A (m2) 0.293296 0.273948 0.264448 0.252848 0.241848 0.236048 0.225298 0.222048 0.222848 0.228348 0.228348 I (m4) 0.246 0.588 1.153 1.848 2.463 2.863 3.486 3.758 4.220 4.609 4.696 IW (m4) 14.4743 9.4030 7.4401 5.6925 4.9346 3.7371 2.6331 2.0441 1.6443 1.4246 1.3515

Fig. 4. Design drawings. Image reprinted from Seyrig [1].

482

A.P. Thrall et al. / Engineering Structures 40 (2012) 479486 Table 2 Material properties. This table provides the assumed material properties include the Modulus of Elasticity (E), the allowable stress (r), and the density (q). Property E

could have been solved for directly [9]. This direct approach is used in Eiffels discussion of his calculations for the Garabit Viaduct [10].

5. Analysis under self-weight, live, and wind load This paper evaluates the behavior of the Maria Pia Bridge under self-weight, three static live loads, and lateral wind load. Fig. 5 provides an elevation drawing of the bridge including numbering schemes for the piers and the nodes of the arch. Note that the intersection of the deck with the arch has been modeled as two piers closest to the crown. Table 2 provides the material properties employed for these analyses. The reader is referred to Table 1 for the properties of the arch given by Seyrig [1]. Seyrig did not provide geometric properties for the piers. We assumed a reasonable value for the cross-sectional area (0.0280 m2) that is concentrated on the edges of the pier (assumed to be constant for its entire length). Based on plans for the bridge, Piers 3 and 4 range in width (in the plane of the bridge) linearly from 4.8 m at the base to 1.532 m at the top [11]. To simplify the piers, we considered only nine sections, each assumed to have a constant width. See Table 3 for the moments of inertia calculated for these constant cross-section sections. Piers 3 and 4 range from 0 m at the base to 42 m at the top. Piers 1, 2, 5, and the piers connecting the deck to the arch each start at a higher elevation, but are assumed to have the same section properties according to their relative locations. Piers 15 are assumed to be xed to the ground. Piers 1, 2, and 5 have roller connections to the deck. Piers 3 and 4 are pin connected to the deck [11]. The piers connecting the arch to the deck are assumed to have a xed connection at the arch and a roller connection at the deck. Seyrig also did not provide any details related to the cross-sectional area and moment of inertia of the deck. To nd reasonable values for these properties, we applied an asymmetrical live load (chosen because it presents an extreme in deck-arch interaction) to a nite element model of the bridge with estimates of deck properties. These estimates were rened until our results matched closely those presented by Seyrig. The cross-sectional area and moment of inertia were chosen to be 0.0326 m2 and 0.1000 m4, respectively. All analyses presented in this research were performed using Structural Analysis Program (SAP). This study assumed an allowable stress of 6000 t/m2 based on the value that Eiffel used in his paper on the Garabit Viaduct [10]. Note that throughout the text,

r q

16 Mt/m2 6000 t/m2 7.8 t/m3

Table 3 Geometric properties of piers. The rst column indicates the height (measured from the height of the arch hinges) corresponding to the assumed section properties. The second column provides the assumed moment of inertia in the plane of the arch (I). Range (m) 05 510 1015 1520 2025 2530 3035 3540 4042 I (m4) 0.149 0.125 0.103 0.084 0.066 0.050 0.037 0.026 0.018

the authors employ the units used by G. Eiffel and Cie. at the time of the design (e.g. tons as the unit of force). 5.1. Self-weight analysis An analysis of the arch under self-weight was performed by building a two-dimensional model of the neutral ber of the arch alone (neglecting the piers and the deck). Hinges on either side of the arch were modeled as pin constraints in the plane of the arch. Table 1 provides the geometry and section properties of the arch. Data related to the lateral bracing was not available. Therefore, to capture the full self-weight of the arch, we applied point loads based on Seyrigs calculations of the self-weight at each node rather than relying on a self-weight load case in SAP. These point loads were back-calculated from the values for moment that Seyrig

Fig. 5. Elevation of bridge with numbering scheme for piers (top) and numbering scheme for nodes of arch (bottom).

A.P. Thrall et al. / Engineering Structures 40 (2012) 479486

483

provided in his numerical integration tables for calculating the horizontal thrust under self-weight only [8]. When analyzed, this model found a horizontal reaction negligibly different from Seyrigs value. The stresses of the extreme ber of the upper chord of the arch are on average 4.6% different from Seyrigs values. At the extreme ber of the lower chord, the average difference is just 1.3%. These comparisons provided condence in our modeling strategy. An efcient two hinge arch design has minimal bending under self-weight. A consideration of the magnitude of bending under this loading will therefore begin our evaluation of the efciency of this design. To visualize the amount of bending, we superimposed the shape of the arch with the magnitude of bending in Fig. 6. The solid line of Fig. 6 shows the shape of the arch. The dotted line shows the moment (divided by 100 for scaling) added to the vertical coordinate of the arch. When the moment is zero, the plots exactly superimpose. A positive moment (above the arch line) implies compression in the top of the element; a negative moment (below the arch line) implies compression in the bottom of the element. Overall the bending moments are small. For example, the highest bending moment occurs at node ve, where the stress from bending is 600 t/m2 at the upper chord and 570 t/m2 at the lower chord. The stress due to bending even at the largest point of moment is just 10% of the assumed allowable stress of 6000 t/ m2, thereby indicating low bending stresses under self-weight.

This study shows that G. Eiffel and Cie. chose a form that minimized bending under self-weight [12]. 5.2. Live load analysis The original live load calculations considered three static loadings: (A) uniformly distributed over the entire arch, (B) distributed over one half of the arch, and (C) distributed over the middle 80 m of the arch with a loading of 4 t/m for each case (see Fig. 7) [1]. We applied these same loadings (assuming the same magnitude) in our full SAP model. Fig. 8 shows the bending moment envelope along the arch for all three loadings. From 0 to 70 m along this diagram, the maximum and minimum moments are given by Load B (considering both halves of the arch). Load B is equivalent to loading the arch at its quarterpoint and therefore, one would expect the highest bending to occur at the quarterpoint as it does in Fig. 8. The spike around 25 m can be attributed to the location where the pier intersects the arch. The minimum moment along the rest of the arch is also from Load B. The maximum moment from 70 m to 80 m is given by Load C. This distribution of the live load bending moment envelope demonstrates the reasoning behind selecting a crescent form for the arch. 5.3. Wind analysis We applied a static, lateral wind load on the arch of the bridge alone. The out-of-plane moment of inertia values for each section of the arch were provided by Seyrig and are given here in Table 1 [1]. Seyrig assumed a maximum wind load of 275 t/m2 which corresponds to a wind speed between 37.3 m/s and 47.4 m/s. However, these high wind speeds would not likely occur over the entire structure at all times while a train was also crossing. Instead, Seyrig assumed a wind load of 150 t/m2 with the presence of a train [1]. In the efciency results discussed in the following section, we considered two, unfactored load combinations: (1) self-weight and 275 t/m2 wind load and (2) self-weight, live load, and 150 t/m2 wind load. These lateral loads are applied as point loads at each node of the arch using magnitudes provided by Seyrig [1]. To understand the behavior of the arch under lateral wind load, Fig. 9 shows the bending moment of the arch under the 275 t/m2 load in the lateral direction. The arch is shown in plan below the

Moment (tm/100) added to Height of Arch (m)

40 30 20 10 0 -100
Magnitude of Moment at Node 5

Arch Moment

-50

50

100

Horizontal Position (m)

Fig. 6. Moment under self-weight (dotted line, quantity divided by 100) added to the geometry of the arch (solid line). The distance between the two lines indicates the magnitude of the bending moment at that node. The lines coincide when there is zero moment.

(A)

(B)

(C)

Fig. 7. Live loadings. Diagram (A) shows the live load applied to the entire arch, (B) shows the live load over one half of the arch, and (C) shows the live load over the middle 80m of the arch.

484

A.P. Thrall et al. / Engineering Structures 40 (2012) 479486

bending moment envelope, and that in plan the arch follows the bending moment diagram under lateral wind load. These studies therefore suggest that the designers at G. Eiffel and Cie. selected an appropriate form based in structural logic. Another method for evaluating the form of the bridge is to consider its efciency (dened as the stress in the member divided by the allowable stress). Though buckling typically controls the capacity for compression members, the slenderness ratio of the most highly stressed member (near the hinge of the upper chord) is 29.5 [13]. As a result, the Euler buckling stress is so high (5,360,000 t/m2) compared to the assumed allowable stress that buckling could not control the design of these members. Therefore, the stress is calculated simply by dividing the axial force by the cross-sectional area. We calculated the efciency of each member of the upper and lower chord of the arch for two unfactored load combinations (1) self-weight and 275 t/m2 wind load and (2) self-weight, live load (worst scenario of three live loadings), and 150 t/m2 wind load. Table 4 provides these calculated efciencies for both load combinations. Fig. 10 shows a visual representation of the efciencies of the upper and lower chords under Load Combination 2. Based on these studies, it is clear that the arch was designed to be very efcient while remaining within the safety limits of the assumed allowable stress.
Fig. 8. Bending moment envelope under three static live loadings with elevation of arch for reference.

6. Elegance of design The elegance of a structure can be considered through three perspectives: (1) the aesthetic motivation of the designer, (2) international acclaim, and (3) an on-site visual analysis.
Table 4 Efciency (e) in the upper and lower chord of the arch (denoted by subscript U and L, respectively) under Load Combinations 1 and 2. Node Combination 1 Combination 2

plot. This shows the highest bending moment occurring at the hinge of the arch and the lowest moment occurring toward the crown. The shape of the arch in plan widest at the hinge and narrowest at the crown therefore follows the bending moment diagram of the arch under wind load. 5.4. Evaluation of efciency of the form Through the analyses described in the previous sections, we have shown that there is minimal bending in the arch under selfweight, that the shape of the arch in elevation follows the live load

eU
1 2 3 4 5 6 7 8 9 10 11 0.729 0.664 0.628 0.567 0.566 0.457 0.432 0.407 0.349 0.338 0.304

eL
0.493 0.513 0.461 0.421 0.357 0.347 0.281 0.302 0.433 0.429 0.466

eU
0.566 0.703 0.801 0.711 0.743 0.556 0.506 0.551 0.259 0.302 0.492

eL
0.741 0.826 0.779 0.752 0.666 0.643 0.523 0.476 0.777 0.785 0.482

0.8 0.7 0.6 0.5 0.4 0.3

Fig. 9. Bending moment under lateral wind load with plan of arch for reference. Fig. 10. Efciency of the arch under Load Combination 2.

C L

A.P. Thrall et al. / Engineering Structures 40 (2012) 479486

485

The aesthetic motivation of the designer was clearly described in Seyrigs report. Seyrig argued that engineers too quickly tend to neglect the study of form. He went on to comment that the desire to provide economic savings often overrides the desire to produce an elegant form. Seyrigs rules of taste, requires that one not lose sight of the fact that the aesthetic considerations are as important as the economic considerations when designing a structure. The Maria Pia Bridge managed to combine economy with pure elegance. Seyrig even wrote that the choice of form for this bridge gives the impression of rigidity and power while maintaining the lightness possible from metal construction. Furthermore, Seyrig makes note that no extra material was wasted for construction or ornamentation, suggesting a focus on economy of materials [1]. Indeed, the form is visually superior to the alternatives presented in the design competition. The next lowest cost design (Fig. 2D) is visually awkward the slender struts supporting the central span give the appearance of a broken arch and the massive abutments create an unpleasant contrast with the delicate metalwork. This further indicates the balance of economy and elegance that the G. Eiffel and Cie. designers found. The bridge has received much international acclaim, including being awarded the distinction of becoming the 11th International Historic Civil Engineering Landmark by the American Society of Civil Engineers (ASCE) in 1990. This is Eiffels only bridge to receive this award [14]. In the nomination for this distinction, Maria Manuela dos Reis Martins of the Universidade do Minho Unidade de Arqueologia wrote With such a remarkable building Eiffel brought about new technical teachings to the world [of] civil engineering and encouraged several engineerings [sic] to set up similar building processes in many other engineering works, mainly in what concerns the crossing of wide deep valleys [15]. A further testament to the signicance of the bridge is Eiffels later masterpiece the Garabit Viaduct (completed in 1884). G. Eiffel and Cie. was given the contract for this bridge without the traditional design competition solely based on the success of the Maria Pia [2,16]. Finally, an on-site visual analysis of the structure can indicate its elegance. The Maria Pia appears to embrace its surroundings as one approaches the bridge. It does not create a disturbance through the ostentation of the Bon March, but rather it is through the simplicity of its design that this iron form appears welcomed in the natural setting of the Douro Valley. The appearance of lightness found in the iron lattice-work is a counterpoint to the rough, dense and rocky terrain of the river and valley that the bridge spans [1]. At the same time, the iron span appears to spring from the earth below. The hinged supports are articulated above the massive masonry structure to create an instant understanding of the purpose of these supports: to concentrate at a single point the weight of the whole. No matter how one approaches the bridge, be it by boat, car or on foot, the Maria Pia does not disturb the landscape. It serves only to add to the charm of this old European city. The surrounding area on the Porto side of the bridge is now lled with shanties, while a funicular runs adjacent to this settlement, up the side of the hill. The winding walk through this make-shift village affords the onlooker occasional views of the bridge, seemingly telling a story of this once vibrant commercial area through its material, span, and structure. The absence of an in-river pier provides a clear path for navigation down the river and allows onlookers the chance to watch as luxury ships travel down the river. The old boats of the various Port-houses line both the Porto and Vila Nova de Gaia riverbanks, re-emphasizing the historical trade in wine on which the city was once dependent. Approaching the bridge from the South via boat, the viewer sees the So Joo Bridge the Maria Pias concrete rigid frame replacement span the river in a greater leap, its strength evident in its slim deck design. Traveling under the bridge,

the viewer is able to fully see its arc, its seeming lightness of structure and the train tracks which had at one point supported the rail trafc between the two sides of the river down to Lisbon and beyond. The Maria Pia with its 1870s era iron structure, latticed piers and deck were to the viewer reminiscent of what would eventually be one of Eiffels most memorable works, his namesake Tower. The Maria Pia gives expression in its design to the difculties it had to overcome in order for it to be constructed. The stone piers on either side of the river enclose the arch as it comes to its hinged support, rising 15 m above the shore-line. This can be attributed to the repetitive ooding of the Douro River, often times rising above the rst oor of the buildings along the river bank. The nished structure shows, in some part, the circumstances and conditions of its construction. The Maria Pia invites contemplation of the evolution of structure and a historical point in time for this beautiful valley rich in port wine and Portuguese literary culture. 7. Relevance to today This paper has examined the Maria Pia Bridge as a signicant structure through its economy, efciency, and elegance. However, the purpose of this paper is not just a case study of a 19th century bridge. Rather, as scholars it is our duty to the profession to present technical, historical, and aesthetic studies of great works worthy of awards like the ASCE International Historic Civil Engineering Landmark distinction which have never been fully documented in civil engineering literature. From such studies, 21st century engineers can understand how a design-build mentality, which characterized the engineers of G. Eiffel and Cie., can lead to signicant works today. Beginning with the industrial revolution, there has been a tradition of great engineers performing planar analyses and using a design-build mentality to develop economic designs. Thomas Telford experimented with arches and suspension bridges in iron. John Roebling pushed the limit of suspension bridge design with his Niagara River Bridge (1855), Cincinnati Bridge (1866), and Brooklyn Bridge (1883) [17]. Robert Maillart designed concrete arches, evolving from his Stauffacher Bridge design (1899) which copied stone arch forms using concrete as the material to the Salginatobel Bridge (1930) which featured a box section design [18]. Othmar Ammann simultaneously built the worlds longest spanning arch (the Bayonne Bridge) and the worlds longest bridge (the George Washington Bridge) in 1931 [17]. Each has used different materials and different forms to generate major works. Each designed a major work which has received the distinction of being an ASCE Historic Civil Engineering Landmark [14]. The common thread among these designers is their conceptual design process which focused on planar analysis and an emphasis on erection process. Engineers today can keep these lessons in mind when developing conceptual designs so that owners and designers can recognize the potential for visually signicant and constructionally economic bridges in the future. 8. Conclusion This paper has discussed the economy of the bridge based on the design competition that earned G. Eiffel and Cie. the contract. Analyses of the structure under self-weight, live, and wind loads have demonstrated its safety and efciency. Within this discipline of efciency and economy, the bridge also exhibits elegance as shown through the aesthetic motivation of the designer, international acclaim, and by an on-site visual analysis. Based on these qualications, the Maria Pia Bridge can be deemed a major work of structural art as it adheres to the following tenets: efciency

486

A.P. Thrall et al. / Engineering Structures 40 (2012) 479486

(minimum material consistent with satisfactory performance and assured safety), economy (competitive construction cost consistent with minimal maintenance requirements), and elegance (aesthetically striking consistent with efciency and economy). These qualications act as scholarly guidelines through which a discussion can be framed. Structural art is a continuum on which designs can be judged, with major works following all three guidelines. All bridges designed with aesthetic motivation and with an efcient form can be structural art. It is the degree to which they are economic and elegant that makes certain bridges major works. Christian Menn wrote that The optimization of economy and elegance requires more than the craftsmanship component of engineering. It requires creativity, fantasy, and sensitivity to visual form. These talents collectively constitute the art of engineering [19]. Here Menn identies the key to great structural art in the engineers play between economy and elegance within a discipline of proper form. This paper has shown that the Maria Pia Bridge meets this denition of a major work of structural art. Furthermore, this paper has demonstrated how developing the conceptual design with the construction process, site constraints, and forces acting on the structure in mind was critical to the success of the engineers at G. Eiffel and Cie. This approach could be used more often by designers and owners today. The Maria Pia Bridge was continually in service from the date of its completion until 1991 when its use was replaced by the So Joo Bridge. After this, it was closed to service, but still stands today [4]. The bridge remains a beautiful reminder of the regions tradition of large span efciently designed bridges. Acknowledgements This material is based upon work supported by the National Science Foundation Graduate Research Fellowship under Grant No. DGE-0646086. The authors are also grateful for nancial support from the National Science Foundation Grant No. 0308549, the Princeton University Gordon Y.S. Wu Chair, the Princeton University School of Engineering and Applied Sciences, the Princeton University Department of Civil and Environmental Engineering, and the Norman J. Sollenberger Fellowship. The authors would also

like to thank the staff at the Archives Nationales du Monde du Travail and the cole Nationales des Ponts et Chausses for their assistance. Finally, the authors are grateful for the advice and guidance of Ted Zoli of HNTB Corporation. References
[1] Seyrig T. Le Pont sur le Douro Porto. Paris (France): Capiomont and Renault; 1878. [2] Loyrette H. Gustave Eiffel. New York (NY): Rizzoli; 1985. [3] Marrey B. The extraordinary life and work of monsieur Gustave Eiffel the engineer. Paris (France): Graphite; 1984. [4] Cruz PJ, Cordeiro JL. Audacious and elegant 19th century Porto Bridges. Pract Periodical Struct Des Construct 2003;8(4):21725. [5] Eiffel G. Notice sur le Pont du Douro, Porto. Clichy (France): Imprimerie Administrative et des Chemins de Fer de Paul Dupont; 1879. [6] Kurrer KE, Kahlow A. Arch and vault from 1800 to 1864. In: Arch Bridges: history, analysis, assessment, maintenance, and repair. Brookeld (VT): A.A. Balkema; 1998. [7] Winter G, Urquhart LC, ORourke CE, Nilson AH. Design of concrete structures. 7th ed. New York (NY): McGraw-Hill Book Company; 1964. [8] Thrall AP. A comparison of the work of Gustave Eiffel and Othmar Ammann: the Maria Pia and Bayonne bridges. MSE thesis; Princeton (NJ): Princeton University; 2008. [9] Parker T. The Douro Bridge. BS thesis; Cambridge (MA): MIT; 1881. [10] Eiffel, G. Mmoire Prsent LAppui du Projet Dnitif du Viaduc de Garabit. Paris (France): Librarie Polytechnique, Baudry et Cie; Republished: Spain: Insituto Tecnico de Materials y Construcciones; 1988 ed.; 1889. [11] Eiffel G. Plans for Pont sur le Douro prs Porto. c.1877a. Box 152 AQ-147, Archives Nationales du Monde du Travail, Roubaix, France. [12] Thrall AP, Billington DP. A study of the evolution of arch forms: Eiffels Maria Pia Bridge and Ammanns Bayonne Bridge. In: Safety and reliability of bridge structures. New York City (NY): New York City Bridge Conference; 2009. [13] Eiffel G. Mmoire lappui du Projet du Pont sur le Douro prs Porto. c.1877b. Box 152 AQ-147, Archives Nationales du Monde du Travail, Roubaix, France. [14] ASCE. Designated historic civil engineering landmarks, <http://www.asce.org/ history/landmark/projects.cfm>; 2009 [Retrieved 12.12.09]. [15] dos Reis Martins MM. International historic civil engineering landmark nomination form 1989. In: Reprinted in Appendix B of: Bra KL. The Ponte Maria Pia: Structural art achieved through purely technical engineering. BSE thesis, Princeton (NJ): Princeton University; 2005. [16] Eiffel G. Notice sur le Viaduc de Garabit. Paris (France): Imprimerie Administrative et des Chemins de Fer de Paul Dupont; 1888. [17] Billington DP. The tower and the bridge. Princeton (NJ): Princeton University Press; 1983. [18] Billington DP. Robert Maillarts bridges. Princeton (NJ): Princeton University Press; 1979. [19] Menn C. The place of aesthetics in bridge design. Struct Eng Int 1996;6(2):93.