Cooling tower structural failures have recently become a focus area for the nuclear industry based on events that have resulted in lost generation as well as high repair costs and personal injury [1].

Besides, cooling tower structures similar to those used in nuclear plants have collapsed at fossil plants. Environmental concerns have also recently increased the need for guidance concerning cooling tower inspection and maintenance. Although no natural draft towers have collapsed at any nuclear power plants, such towers have suffered collapse throughout the world. Notable collapses are as follows:

  • In 1965, three out of a group of eight reinforced concrete natural-draft cooling towers were blown down in strong winds at Ferrybridge Power Station in the United Kingdom
  • In 1973, a single 137-metre-tall natural-draft cooling tower collapsed under moderate winds at Adeer Nylon Works power plant just off the southwestern coast of Scotland
  • In 1979, one natural-draft tower at Bouchain in France collapsed under minimal winds. This tower was known to have had serious dimensional errors from the beginning, and it is now suspected that the collapse could have been caused by progressive deterioration [2]
  • In 1984, one natural draft tower under construction at the Fiddlers Ferry Power Station collapsed under high gusts of wind in the UK.

In France, EDF operates a fleet of 28 reinforced concrete towers built from 1977 to the middle of the 1990s at nine of its 19 nuclear power plants (NPP). These structures are not safety-related, but considering the investment and their role in the plant operation, EDF aims to anticipate the effects of the ageing phenomena. Since the beginning of operational life, structural surveys comprising in-service inspection and monitoring have been implemented. A standard in-service inspection programme implemented for each tower includes visual examination, topographical measurements and surface mapping of the shell. In case of observed significant distress, specific analysis on core samples from the shell or by non destructive evaluation could be carried out. Monitoring outputs feed databases which are used to compare, rank and build a maintenance plan across the fleet.

Design basis and ageing

Contrary to popular belief, the shape of a hyperbolic natural draft cooling tower has nothing to do with the airflow through it. The airflow through the tower is generated by the difference in density between the warm, humid air inside the tower and the relatively cool, dense ambient air outside the tower. Originally, natural draft cooling towers were cylindrical in shape. As the design of these types of towers evolved, and the towers were made increasingly larger, the cylindrical shape was changed to hyperbolic, which offers superior structural resistance to ambient wind loadings with an optimized volume of concrete [3].

EDF’s standards for reinforced concrete cooling towers require a minimum duration of 30 years for operations [4]. The lessons learnt from collapses or heavily-damaged towers in the years 1960-1980 are taken into account in these technical specifications. Reinforced concrete cooling towers may be subjected to a variety of loading conditions. Most commonly, these are dead load, wind load, earthquake load, temperature variations, construction loads, and settlement. That standard comprises requirements on:

  • Shell wall thickness: no less than 16 cm, in order to enhance durability of the concrete
  • Reinforcement: a mesh of two orthogonal layers of reinforcement should be provided in the shell walls, generally in the meridional and circumferential directions
  • Non-uniform settlement effects
  • Temperature effects: uniform, gradient through the wall, sunshine on one side
  • Concrete cover: no less than 3 cm, to preclude rebar corrosion
  • Stiffening of the upper and lower edge of the shell.

Modern cooling towers can reach 180 m high with a diameter at the base of about 140 m. The thickness at the throat level is only about 20 cm, which is why it has often been claimed that this kind of building is thinner than an eggshell. The tower is supported by a large number of columns (vertical, diagonal, or cross-shaped) linked together by annular footing which transmits loads to the ground (Figure 2). In a standard zone of the shell, the steel ratio is about 0.4-0.5%, placed at the inner and the outer face.

It is well-known that concrete structures are not built to last forever. In normal exposure conditions, they deteriorate over time. A Cooling tower’s environment can speed up expected ageing phenomena or entail serious damage that would not occur otherwise. Overviews of the main degradation mechanisms of concrete cooling towers are periodically provided [6, 7]. The most frequently-addressed phenomena are thermal and moisture through-wall gradients, freeze-thaw, ice formation, rebar corrosion (due to chloride ingress or carbonation) and alkali-silica reaction.

As a matter of fact, all of these degradation mechanisms are more or less linked with cracking and rebar corrosion. Rebar corrosion has been identified by EDF’s researchers and engineers as the main threat for cooling tower integrity. Decrease of the overall bearing capacity is not the only concern. Extensive cracking or spalling can also result in concrete blocks falling, damaging surrounding buildings or equipment (for instance, distribution basins or canopy beams), or posing dangers for people working nearby.

Corrosion of embedded steel reinforcement can be induced by several factors. Normally, concrete provides a high alkaline environment (pH around 12.5) and a thin passivating film forms on the steel, preventing it from corrosion. In case of chloride ingress or carbonation of concrete, the protective coating can be compromised by a pH drop.

Up until a threshold concentration is reached, chloride ions can chemically bind with cement constituents. Above this threshold, in the presence of oxygen and moisture, chloride can penetrate the passive film and lead to local corrosion of steel rebar. The rate of chloride ingress will depend on concrete properties (particularly permeability and chloride sorption capacity) but also on cracking. In presence of open cracks, access to uncoated steel will be easier and its corrosion faster.

Carbonation is experienced by almost all concrete structures, but its depth and rate depend on a few specific factors. They are:

  • the air’s permeation into the concrete
  • the level of atmospheric carbon dioxide
  • the moisture content of the concrete (mean internal humidity and variations due to operations and weather changes).

Like other nuclear power plant operators, EDF undertakes preventive and long-term maintenance for all of the civil engineering structures relevant for safety and for power generation. The objective of routine maintenance is to preclude any major expensive, disruptive and time-consuming repairs. Standard guidelines and maintenance rules are periodically applied for all the cooling towers. Relying on collected data, risk analyses are undertaken to determine if repairs are necessary, and if so whether repairs or retrofits would be the most appropriate.

If the owner aims to extend the service life of the cooling towers, more extensive investigations and thorough analysis can be performed to determine the remaining operating life and the current safety margins. Nevertheless, there is currently no generically-accepted method for cooling tower service life assessment. The opinions of in-house experts and external consultants can help managers and investors determine whether a new tower or major repair will be required within a period of 5 to 10 years.

EDF ageing management is mainly based on IAEA Safety Guide

NS-G-2.12 [8]. This standard is not specific to civil structures but it provides a general methodology to address ageing issues. Monitoring is clearly identified in the overall organization for implementing efficient ageing management. According to the same principles, EPRI and EDF have issued an ‘end user’s flowchart’ for lifetime management [9]. Such an assessment must utilize a cross-disciplinary analysis that takes advantage of knowledge of materials science, non destructive testing, visual inspection and, where appropriate, computational mechanics (Figure 3).

Surveillance programme

Although cooling towers are not safety-related, EDF decided to start inspecting all of them and carry out measurements of the real shape of their outer surface by geodetic and photogrammetric techniques. Subsequent inspections were carried out in the aim of mapping all defects and deformations and to record and follow up the adverse ageing effects. EDF follows a detailed procedure for these inspections, involving:

  • measuring the differential settlement and potential tilting of the structure using traditional topographical equipment
  • performing a detailed visual inspection of the concrete surface by mapping the shape and size of the cracks
  • mapping of the real shape of the shell by planimetry at different levels and by photogrammetric measurements or a 3D laser scan survey.

Vertical displacements of the bearing columns had been measured by topographical levelling since the beginning of the erection of the towers. A network of markers fixed on the circular ring footing enables a periodic monitoring with a levelling instrument. EDF decided to carry on with this survey also during cooling tower operation, to check that observed settlement does not exceed design hypotheses. Simple movement plotting over time is useful to detect any local or global unexpected or abnormal trend in settlement (Figure 4). The frequency of these surveys is adapted to foreseeable and observed movements, between 1 and 4 years. Measurement uncertainty is about 1 mm. Generally, observed settlements are mostly stabilized within 10 years after the end of construction.

However, other situations may be encountered. It can be the case, for example, at an upward soil swell occurs when water leaks through cracks in the cold water basin. The swell can entail upward displacements of footings and supporting columns (Figure 5).

Moreover, topographical surveys are carried out to monitor horizontal displacements, benchmarks being set at several levels of the shell and periodically surveyed with a theodolite. These planimetry measurements are combined with those from tower geometry surveys (see below) to assess the overall deformation of the tower.

A visual inspection is a thorough examination of the concrete to identify its condition and the possible outward signs of ageing that may be encountered during the service life of a building. Requirements for in-service inspection of nuclear civil structures entail visual inspection of visible concrete parts [10].

A convenient way to undertake reliable visual inspection for ageing management consists of searching for categorized patterns, distresses and imperfections with known features. For example, IAEA and the ACI proposed some guidance to support such a survey [11, 12]. EDF has issued its own specifications, with standardized checklists, output formats and reports. For example, a crack is documented by its width, length, direction and pattern. Distresses and defects linked with corrosion marks are carefully mapped. When a defect is recorded, it is stored in a database with its characteristics with at least one picture.

For simple buildings, a visual inspection can be undertaken with rather simple devices, but for structures like cooling towers, surveyors need remote systems. The surface of the shell is scanned by a long focal length telescope coupled to a video camera driven by an accurate geodesic system to acquire reliable data on defects positions and features. An image storage system is used to record the results. These devices are able to estimate crack length and opening with a 0.1 mm resolution (Figure 6).

The frequency of visual inspection of cooling towers ranges between 3 and 6 years. For each campaign, a final report provides an overview of the observed ageing patterns. For example, the number of cracks, corrosion marks or spalling are computed, statistical analyses are performed for different areas of the shells and comparisons with previous inspections are made to assess ageing kinetics.

If visual inspection and foundation displacement seem to be included in most of the routine inspection guidelines, shell shape monitoring is required in some countries which have experienced failures due to huge geometrical imperfections, as reported in [13]. EDF’s practice includes such measurement and analysis. Some towers in France have been demolished in the past for distortion that was so excessive it posed a high risk of collapse (at the Pont-sur-Sambre and Ansereuilles fossil power stations).

Until recently, photogrammetry was the main technique used to monitor the real shape of the outer surface of the cooling towers and to record geometrical imperfections. Photogrammetry is a practice of determining 3D coordinates of points on an object from photographic images. These are determined by measurements made in two or more photographic images taken from different positions by calibrated cameras. The geometric reconstruction is derived from triangulation of several common points identified on each image.

A photogrammetric survey performed for a natural-draft cooling tower on EDF’s nuclear site involves a network of 16 camera stands surrounding the structure situated at a distance between 50 and 100 metres of the footing. Sophisticated algorithms and software enable the integration of other data such as theoretical geometry of the objects. The deviation between the theoretical and the measured shell shapes can be analyzed and followed up over time. Combining these results with those of a classical planimetry survey enables greater accuracy of the measurements.

Nowadays, photogrammetry has been replaced by 3D scanning surveys. EDF has chosen laser devices that collect positioning data as point clouds, using reference markers but also natural features on the structures (Figure 1). These techniques enable a determination of an almost unlimited number of points, which entail a more accurate description of the cooling towers (Figures 7&8). Moreover, acquisition and processing are less time-consuming than photogrammetric surveys.

The frequency of shell shape surveys ranges between 4 and 6 years. Uncertainty measurement is evaluated between 10 and 30 millimetres at the top of the structure, taking into account the performance of the devices themselves but also the layout of the benchmarks and the distance from the stations to the structure.

A summary report is issued every year for the entire cooling tower fleet. It gathers together all the main outputs of previous measurements and inspections, presented on a single sheet for each tower. The document concludes with a ranking of all the towers. This ranking is based on experts’ judgments on whether some towers need further investigation (for example, local non-destructive test for corrosion detection, core sampling for laboratory analysis) or if repairs are to be planned within the next few years.

Ongoing development

It has been claimed that the current methodology to determine the priority for detailed analysis or for strengthening repairs is based on experts’ opinions. Without questioning the skills of its board of experts, EDF decided to test other options to assess tower degradation. One of the difficulties in analyzing the data arises from the heterogeneous nature of the source material, which consists of pictures, numbers, more-or-less accurate measurements, or from data based on their distribution in time and in space. Symbolic data analysis (SAD) techniques enable the extraction of knowledge from these heterogeneous-object databases and to summarize the large data sets in such a way that the resulting database is of manageable size. A way of doing this kind of information clustering is to change the standard format of data: the collected data are represented by lists, intervals, distributions and the like. These summarized data are examples of symbolic data, which can then feed more classical statistical tools.

A first attempt to apply SAD to cooling towers was carried out under the umbrella of a national project dedicated to structural health monitoring for civil structures, supported by the French sustainable development ministry. Indeed, in this project the towers were described in a single table merging all monitoring measurements into a set of variables from different kinds of data (continuous, nominal-valued, interval or histogram-valued variables). Using this master table, a linear statistical model mixing these degradation variables has been performed to rank the towers according to maintenance priorities.

It was interesting to note that the linear model gave the same ranking as the experts, except for one tower. The SAD-determined ranking of this tower indicated greater deterioration than the experts’ conclusions. This discrepancy is explained by the fact that the initial shell shape survey for this tower did not follow the standard procedure. Due to the uncertainty assessment for the initial measurements, experts tended to minimize the weighting of geometry control, a detail which was overlooked in the automatic symbolic data processing. An illustration of the capacities of the software implemented for EDF’s cooling towers ageing analysis and some results were presented in [14].

The current surveillance system for EDF’s cooling tower shells is considered to be reliable and robust, but only for the purpose for which it has been designed. Nowadays, the extensive use of finite element modeling to analyse adverse ageing effects on structural behaviour makes new demands on monitoring. For example, engineers would like to know how a tower shape is modified by the effect of a day’s sunshine or by a plant startup in freezing winter conditions. They are also looking for more information on the temperature and moisture gradients through the concrete wall of the shell too.

Embedded temperature or strain sensors appear to be a convenient way to monitor the structure for such aims. One tower in EDF’s fleet has been equipped by these kind of sensors embedded in its concrete wall (temperature, humidity probes or vibrating wire strain gauges). These sensors can help to better understand concrete behaviour during erection and in operating conditions. For example, [15] presented an analysis of data provided by vibrating wire strain gages, focusing on the hydro-mechanical couplings in the concrete shell. If an expected shrinkage of concrete was observed during construction, the start of the plant entailed a high level of moisture released on the inner surface of the shell and then experienced concrete swelling by free water ingress. This kind of device is accurate and robust. In the field of dam monitoring, EDF has witnessed some vibrating wire strain gauges continuing to function more than 50 years after the start of operation. EDF intends to require such instrumentation for future towers, to better anticipate possible delayed adverse ageing phenomenon.

Where no sensors have been embedded in the tower’s concrete, several options will be assessed by EDF in the coming years:

  • Core drilling of the shell to install new sensors
  • Placement of a large set of new topographical marks within the outer surface of the shell, and 3D displacement monitoring with an automatic theodolite (data acquisition frequency around one hour, over several months);
  • Fibre-optic strain and temperature sensors, promising technologies for civil structures monitoring, mounted in the outer and inner surfaces of the shell.


Alexis Courtois (, Yves Genest (, Anne Vacqué (, EDF DTG, Lyon, France. Filipe Afonso (, Syrokko, Paris, France.

This paper is based on a presentation to the Joint Electric Power Research Institute/EDF/Materials Ageing Institute workshop on inspection and degradation management of concrete structures in the nuclear industry, 14-16 September 2011, Materials Ageing Institute, Moret-sur-Loing, France.

This article was published in the April 2012 issue of Nuclear Engineering International magazine.



[1] Boles G. 2011, Nuclear Maintenance Applications Center: Guideline for Cooling Tower Inspection and Maintenance. Technical Report 2011.1021060. Palo Alto : EPRI.

[2] Courtois A., Barnel N & Ravet S. 2007. A simulation-based method to estimate the ageing of RC cooling towers. In F. Toutlemonde et al. (eds), Concrete under Severe Conditions: Environment & Loading, CONSEC07, Proc. Symp. Tours, France, 4-6 June 2007: 1093-1100, Paris: LCPC.

[3] Caudron L. 1991. Les refrigerants atmospheriques indutriels. Paris: Eyrolles. (in French)

[4] EDF 1991. Ouvrages en baton des refrigerants atmospheriques humides à contre courants et tirage naturel. Clauses Generales. Technical Specifications 56.C.005.00. Paris: EDF.

[5] Kratzig W.B., Petryna Y.S. & Stangenberg F. 2000. Measures of structural damage for global failure analysis. International Journal of Solids and Structures 37: 7393-7407

[6] Guimaraes M. 2010. Program on Technology Innovation: Assessment of needs for concrete research in the Energy Industry. Technical Report 2010.1022373. Palo Alto: EPRI

[7] Salomon M. & Gallias J.-L. 1991. Durabilite des voiles minces en breton arme. Cas des refrigerants atmospheriques. Annales de lTBTP 496: 13-47 (in French)

[8] IAEA 2009. Ageing Management for Nuclear Power Plants. Safety Standards Series no° NS-G-2.12, Vienna: IAEA.

[9] Le-Pape Y., Wall J. & Salin J. 2011, A general stratgey to provide an effective owner-orientated toolbox to support long term operation of civil infrastructure. EDF report H-T25-2009-00877-EN. Moret-sur-Loing: EDF

[10] Naus J.-L. 2009, Inspection of nuclear power plant structures: overview of methods and related applications. Report ORNL/TM-2007/191. Oak Ridge, Tenesse, USA: Oak Ridge National Laboratory.

[11] ACI 2008. Guide for Conducting a Visual Inspection of Concrete in Service. ACI Committee guide ACI 201.1R-08. Farmingtion Hills : American Concrete Institute.

[12] IAEA 2002. Guidebook on non-destructive testing of concrete structures, Training Courses no 17, Vienna: IAEA.

[13] Bamu P.C. & Zingoni A. 2005. Damage, deterioration and the long-term structural performance of cooling-tower shells: A survey of developments over the past 50 years. Engineering Structures 27 (2005): 1794-1800.

[14] Afonso F. et al. 2010. Use of symbolic data analysis for structural health monitoring applications. In 2nd International Symposium on Life-Cycle Civil Engineering, IALCCE2010, Proc. Symp. Taipei, Taiwan, 27-31 October 2010.

[15] Chauvel D. & Barre F. 2005. Concrete delayed deformations of nuclear structures comparison between monitoring analysis and theoretical values. In Pijaudier-Cabot et al. (eds) Creep, Shrinkage and Durability of Concrete and Concrete Structures " CONCREEP-7, Proc Symp. Nantes, France, 12-14 September 2005: 317-322. London: Hermes Science Publishing.