Leaning Tower of St Moritz

The Leaning Tower of St. Moritz: a structure on a creeping landslide

This paper is based on [1]

1  Abstract

The 13thcentury Leaning Tower of St. Moritz is located in the historic centre of the famous Swiss ski resort in the compression zone of a 10-million m3 Brattas-Fullun landslide. Over hundreds of years, this slowly creeping landslide, which is blocked by a rock outcrop below the tower, had damaged the adjacent St. Mauritius church to such an extent that it had to be demolished as early as 1893. The fact that the 33-m tall tower has survived its 5.40-degree downslope inclination should not be taken for granted: this is the outcome of a century-long effort by several outstanding Swiss engineers who came up with original stabilisation solutions over the years. This paper explores the history of the Leaning Tower of St. Moritz, describes various attempts at stabilising it and pays tribute to some brilliant minds behind them.

2  Introduction

The historic Leaning Tower of St. Moritz (Figure 1) is located 50 m uphill from the Kulm Hotel in the centre of the village. All the historic structures above the hotel suffered from slowly accumulating ground displacements and had to be eventually demolished. Except for one – the Leaning Tower of St. Moritz.

Figure 1 The Leaning Tower of St. Moritz (from Oberender/Val/Puzrin)
3  The Brattas landslide

The Brattas-Fullun landslide, which constitutes the primary factor for the special geotechnical conditions of the Leaning Tower, is located on the northern slope above the village of St. Moritz (Sterba et al., 2000 [8]). It consists of a 600-m vast clastic flow bounded by almost parallel shear surfaces (Figure 2, top) on both sides. The detachment zone is located on the southern edge of the terraced surfaces of Val Saluver at an altitude of 2,400 m a.s.l. .The area stretches over a horizontal distance of 1.5 km to a lower altitude of 1,850 m a.s.l with an average inclination of about 20°. This clastic flow consists of two parts (Figure 2, bottom), with some geological evidence of a rock outcrop at the boundary between them. The lower 600 to 700-m long zone, the actual Brattas-Fullun landslide, consists of a 17 to 23-m thick silty soil matrix with boulder inclusions (slope detrital), whose downhill movement is constrained by a rock outcrop at its foot. The landslide area is partially wooded at the top, while its lower portion has been developed for construction purposes. Until 2006, the movement was only measured in developed areas, reaching 5 cm per year above the construction zone, while gradually tapering off to zero near Via Maistra at the lower landslide edge.

Figure 2 The St. Moritz-Brattas-Fullun landslide (left), aerial photo (after Google); (right) the geology (Müller and Messina, 1992).

The large-scale geological situation of the area resulted from Mesozoic sediments of the Bernina Nappe being pushed over the crystalline rock of the Err Nappe. This movement subjected all the layers in the region to high tectonic stresses and formed a lithologically extremely heterogeneous and strongly weathered weak zone with liassic layers of clay-rich schists, which facilitate the formation of slip surfaces.

Specific hydrological conditions further contributed to slope instability. For example, the saturation of the slope is not only due to frequent precipitation (1,000 mm on average per year) but also to permanent seepage from the Schlattain brook and periodic snowmelt from the terraced surfaces of Val Saluver (Figure 2, bottom), verified through tracer tests by Müller and Messina (1992, [6]). Various deep aquifers were observed in the landslide, which created independent water tables in the slope.

4  The Leaning Tower and the stabilisation of the landslide (Albert Heim, 1899)

The bell tower of the St. Mauritius church was initially built in Romanesque style in the 13th century, approximately 100 years after the church had been built (e.g., Haefeli, 1974 [4]). Its height increased over the centuries: In 1890, it was decided to reduce the overturning moment by removing the chimes and contain its growing inclination. In 1893, the church had to be pulled down as differential displacements in the church reached a level that seriously endangered its structural stability. Demolishing the tower was also considered in 1897, but it could remain in its inclined state (estimated at 2.5° − 3.0° at that time, Wullimann and Schneller, 1989 [10]), its masonry being of excellent quality. However, the St. Moritz community was still concerned about continuing displacements in the area and the tower’s inclination: in 1899, it invited Prof. Dr. Albert Heim (1849 – 1937) to investigate potential causes and suggest possible mitigation measures.

In his expert report of 1899, Prof. Heim established that the main reason behind the relentlessly increasing displacements in the Leaning Tower of St. Moritz and surrounding structures was a permanent landslide. Based on this finding, he suggested reducing its velocity by installing a drainage system in the upper part of the landslide. However, it could not entirely stop the displacements, as shown by regular tower inclination measurements conducted from 1908 onwards (Figure 4). By 1928, the inclination of the tower reached 4.4° (7.7%), and the danger of a tower collapse became imminent (Maillart, 1931, [6]).

Figure 3 Albert Heim (source ETH-Library)
Figure 4 Inclination of the tower in % between 1908 and 1967 (after Haefeli, 1974, [4])
5  First attempt at stabilising the tower (Robert Maillart, 1928)

In 1927, the concerned municipality invited Engineer Robert Maillart to explore the possibility of stabilising the tower. Robert Maillart (1872 –1940, Figure 5) was a famous Swiss civil engineer and bridge designer.

For the Leaning Tower, Maillart established that the lower edge of the foundation was experiencing a 700-kPa pressure, which not only contributed to the tower’s inclination, but could also bring it close to a bearing capacity failure (Maillart 1931 [5]). It was decided to implement stabilisation in two steps (Figure 6): (i) decreasing existing earth pressure upslope and (ii) increasing the foundation area downslope.

The task was challenging since it involved excavation into soil and masonry at the overstressed downslope part of the tower. In order to avoid potential collapse while stabilising, the tower was pulled back by 16 tension cables connected to top openings in the tower and anchored back in heavy concrete blocks.

Figure 5 Robert Maillart
Figure 6 The first stabilisation of the Leaning Tower (after Maillart, 1931 [5])
6  Second attempt at stabilising the tower (Robert Haefeli, 1968)

From 1961 onwards, the tower’s inclination accelerated (Figure 4), so the municipality of St. Moritz deemed it necessary to invite an expert opinion – at that time, Professor Robert Haefeli. Robert Haefeli (1898 – 1978, Figure 7) was not only the pioneer of geotechnical research in Switzerland but also one of the founders of the International Society for Soil Mechanics and Foundation Engineering.

He was appointed a professor at ETH Zurich in 1947 and lectured on soil and snow mechanics in combination with avalanche mitigation and preventive measures.

The immense value of his achievements has become more evident against the backdrop of tragically long bouts of depression, which Haefeli suffered over the years and ultimately forced him to leave ETH Zurich in 1953. However, he remained active as a private consultant: In 1967, the St. Moritz municipality called upon his expertise for a next attempt at stabilising the Leaning Tower, the inclination of which reached 5.37° (9.4%) at the time.

Figure 7      Robert Haefeli (after Burland, 2008 [8])

In his assessment of the Leaning Tower conditions, Prof. Haefeli calculated stresses in structure and soil caused by the tower’s inclination and the wind load, which had not been considered before. Since Maillart’s last stabilisation attempt in 1928, stresses in the soil had increased from 200 kPa to 290 kPa but without reaching bearing capacity. Therefore, the main concern was the tower’s increasing rate of incline (from 0.02% per year between 1939 and 1961 to 0.05% per year between 1961 and 1964), which could eventually bring stresses to critical values.

Professor Haefeli suggested stabilising the tower in three stages (Haefeli, 1974, [7]). In Stage I, the foundation was anchored in the soil behind the tower (Figure 8) through four 300 kN, 20 m long Duplex anchors. Stressing the anchors caused a 0.01°- rotation in the upslope direction (opposite the tower’s lean), confirming the solution’s effectiveness. Stage II included installing a drainage system below the tower’s foundation. Three horizontal 25-m long, 131-mm diameter wells spread in a fan were drilled below the tower (Figure 8) to carry out the operation. Although only one pipe carried a few groundwater drops, the measure proved to be rather effective, causing the tower to rotate backwards by 0.03° almost immediately. A possible explanation is that lowering one of the local groundwater horizons caused an increase of effective stresses, resulting in settlements forming behind the tower. The first two stages proved so effective that Stage III could beabandoned. However, it is still worth mentioning to illustrate Prof. Haefeli’s creative thinking. To be able to apply the resisting moment to the tower foundation, he suggested building a lever system [4]. According to Prof. Haefeli’s explanation, applying a moment opposite the overturning moment could not only reduce stresses in soil and structure but also affect the kinematics of tilting by triggering a soil creep in the opposite direction.

Figure 8 The second stabilisation of the Leaning Tower (after Haefeli, 1974 [4]).

7  Third attempt at stabilising the tower (H.-J. Lang, R. Wullmann, F. Schneller, 1983)

Professor Haefeli continued to monitor the tower after the second stabilisation attempt. In 1973, the anchors, which seemed to have lost some force, were pre-stressed back to 300 kN. In 1976, due to health problems, Professor Haefeli had to terminate his consulting activity and the municipality contracted the Institute of Geotechnical Engineering, ETH Zurich (at that time Institut für Grundbau und Bodenmechanik, IGB, headed by Prof. Hans-Jürgen Lang from 1970) to ensure further monitoring of the tower.

Hans-Jürgen Lang (1929-2017) graduated in Civil Engineering at ETH Zurich. After a distinguished career in industry, he was appointed the first professor in soil mechanics and geotechnical engineering at ETH Zurich in 1968.

The sensitivity of the tower to relatively weak seismic loads, like the Friuli earthquake of May 1976 in northern Italy (M=6.5), became another source of concern. Therefore, it was decided to review the tower’s stability, following the results of existing site investigations, i.e. inclinometer and piezometer measurements. Based on limit analysis and finite element calculations, it was concluded that the safety factor against bearing capacity failure below and outside the tower was as low as FS=1.04 and could quickly drop to unity if the groundwater level were to rise to a relatively moderate extent (Wullimann and Schneller, 1989 [10]).

Figure 9 Schematic drawing of the third attempt at stabilising the Leaning Tower of St. Moritz (information poster from the St. Moritz community [1])

In 1982, IGB (R. Wullimann and Prof. H.-J. Lang) and the engineering firm Edy Toscano AG (F. Schneller) (Figure 9) came up with the following solution [10] to stabilise the tower. First, the tower was underpinned by two reinforced concrete foundation walls carefully sunk to 10 m, i.e., about 3 m above the sliding surface. Next, pre-stressed reinforced concrete collars were placed near the foot of the tower (Figure 10) to pre-stress the masonry of the foundation and bind the original foundation to the footing extension constructed by Maillart in 1928. The tower weight (1264 t) was then lifted using hydraulic jacks and transferred onto the foundation walls with three Teflon bearing pads located between the collar beams and the foundation walls (Figure 11). Finally, the tower’s masonry was reinforced using vertical internal pre-stressing through tension cables. Like in 1928, during the first stabilisation attempt, the safety of the tower during construction was ensured through cables and ground anchors, whereby cables were placed in the pipes to reduce temperature strains.

Figure 10    Construction of pre-stressed reinforced concrete collars at the tower footing
Figure 11    Accessible Teflon bearing pad

After construction was completed in 1983, the tower’s lean decreased by about 0.27°, bringing it thirty years back in time. Since that year, the tower inclination’s has been carefully monitored by the St. Moritz municipality (Mr. Pietro Baracchi) using two independent measuring methods: theodolite outside and Zenitlot inside the tower. According to the measurements, the tower approached its 1983 lean (Figure 12) both in 2005 and 2013. It was corrected in 2005 by the engineering firm Edy Toscano AG with ETH Zurich’s support, like in 2013.

Figure 12    Recent measurements of tower deflection from the vertical over its 21.30-m height

8  Monitoring

Today, it is possible to identify the exact cause of the tower’s faster tilting by using displacement measurements. Displacements and rotations of the tower have been repeatedly measured since 1908 (Figure 13).

Figure 13    Approximate historical measurements of tower deflection from the vertical over its 21.30-m height [1]

Since the early 1980s, the tower has been monitored regularly using several additional manual measuring methods. Consistent with other observations, measurements show that in the past three decades, the ground surface in the vicinity of the tower experienced faster average downslope displacement rates (Figure 14): 6.5 mm/year from 1986 to 1996; 12 mm/year in 1996 to 2006 and 19.5 mm/year in 2006 to 2016. However, according to inclinometer readings (Figure 15), displacement rates tend to decrease with depth, reaching about 10 mm/year (from 2006 to 2010) at the slip surface. This non-uniform velocity field is bound to affect the 10-m deep tower foundation walls. At the ground surface, their rotation expresses itself as a differential settlement/heave at the tower’s four corners, leading to faster tower lean. Using a measurement base distance of 21.3 m (Figure 14), the average tower deflection rate of 6 mm/year corresponds to a tilt rate of approximately 0.016°/year or about 3 mm per 10 m of foundation wall depth, consistent with inclinometer readings in Figure 15.

Figure 14    Ground displacements and cumulative inclination of the tower between 1978 and 2016 [1]
Figure 15    Average displacement rates in depth measured with inclinometer near the Leaning Tower between 2006 and 2010 (according to Puzrin and Schmid, 2011)

Acceleration in landslide displacement rates caused by soil failure in the compression zone has been closely followed by the tower inclination (Figure 14). In order to monitor these developments more accurately, react adequately to extreme events, and complete manual measurements, an automatic monitoring system designed by the Institute of Geotechnical Engineering, ETH Zurich, was installed by Solexperts AG. This system includes an inclination sensor, temperature and air pressure sensors, two in-place inclinometers, one piezometer and three accelerometers. All data is accessible online, and automatic text messages are sent when readings reach critical values.

This extremely sensitive system is able to detect tiny changes in tower inclination and pore water pressure caused by construction work in the vicinity of the tower.

9  Outlook

More than a hundred years of efforts by brilliant Swiss engineers have kept the Leaning Tower of St. Moritz alive, long after all other structures in its vicinity were pulled down, rebuilt and demolished again. How to ensure the stability of a tall and narrow historic structure in the compression zone of a creeping landslide is not a skill a geotechnical engineer is taught at school. The vast, 10-million cubic meter landslide is not about to stop; quite the opposite: it is currently progressing while existing Teflon bearing pads have exhausted their capacity to adjust their heights by adding or removing plates. Furthermore, it has been established (Edy Toscano AG) that the existing solution does not satisfy the latest seismic stability standards.

According to stochastic models of landslide displacements (Oberender and Puzrin, 2016, [11]), the tower is set to reach its record 5.40-degree inclination between 2025 and 2030. A new solution must be found and implemented to prevent this, including seismic isolation. Hopefully, the Leaning Tower of St. Moritz will continue to challenge new generations of engineers for many years, decades and centuries to come!

10 References

[1] Alexander M. Puzrin (2017) The Leaning Tower of St. Moritz: A structure on a creeping landslide, Geotechnics and Heritage: Historic Towers, edited by R. Lancellotta, Renato, A. Flora and C. Viggiani, pp. 123-143, Boca Raton: CRC Press, 2017.

[2] Brockmann-Jerosch, M., Heim, A. & Heim, H. (1952) Albert Heim – Leben und Forschung. Wepf & Co., Verlag, Basel.

[3] Haefeli, R. (1944) Zur Erd- und Kriechdruck-Theorie. Schweizerische Bauzeitung, Band 124, pp. 256–260.

[4] Haefeli, R. (1974) Der schiefe Turm von St. Moritz im Vergleich zum schiefen Turm von Pisa. Schweizerische Bauzeitung Band 92, Heft 16, pp. 381-388.

[5] Maillart, R. (1931) Die Erhaltung des schiefen Turmes in St. Moritz. Schweizerische Bauzeitung, Band 97/98, Heft 3, pp. 29-31.
[6] Mueller, E.R. & Messina, G. (1992) Geotechnische Gutachten, Bericht Nr. 2570-1, Rutschung Sass Runzoel-Brattas, St. Moritz. Buechi und Mueller AG, Chur (unpublished)

[7] Schwager, M. & Puzrin, A.M. (2014) Inclinodeformometer for earth pressure measurements in creeping landslides. Géotechnique 64, No. 6, pp. 447-462.

[8] Sterba, I., Lang, H.-J. & Amann, P. (2000) The Brattas Landslide in St. Moritz. Proceedings of GeoEng, Melbourne

[9] Burland, J.B. (2008) The founders of Géotechnique. Géotechnique 58, No. 5, pp. 327-341

[10] Wullimann R. & Schneller, F. (1989) Der schiefe Turm von St. Moritz. Schweizer Ingenieur und Architekt, Band 107, Heft 35, pp. 901-906.

[11] Oberender, P. W. & Puzrin, A. M. (2016). Observation-guided constitutive modelling for creeping landslide, Géotechnique 66, No. 3, 232–247