MONITORING DEFLECTIONS OF HISTORICAL LARGE TIMBER ROOF STRUCTURE USING DRAW-WIRE SENSORS

Maciej A. Orzechowski, Radosław Tatko
Institute of Building Engineering, Wrocław University of Environmental and Life Sciences, Wrocław, Poland

ABSTRACT

This article presents a new approach to monitor structural behavior of large historical timber structures. During the assessment of the method, data was collected at the research station located in the attic of the historical roof structure of a medieval church. The collected data was analyzed in terms of the influence of environmental actions (such as temperate change and wind load) and the results are presented in this article. This project is significant because the measurements of deflections of a very large structure are taken in situ. The article explains the complexity of such measurements as well as provides an overview of selected methods of measurement of structure deformations. A new application of draw-wire sensors as measurement instrument is presented with the results confirming the applicability of such sensors for deflection measurements

Key words: structural health monitoring, draw-wire sensor, environmental actions, timber structure, large structure.

INTRODUCTION

Structural health monitoring is a very important aspect of ensuring the safety of structures. For the modern construction the installation of various monitoring devices is relatively easy, as at the time of designing their location can be carefully planned. However, for existing structure, especially for large  historical structures, the application is not straight forward. This article discusses the preliminary analysis of data recorded at the research station located on the historical church of St. Dorothy, Vaclav and Stanislav in Wroclaw, Poland. The analyzed structure is constructed of timber roof trusses, of a height and span of 24 m, was constructed approx. in 1400 and it is the largest structure of this kind in Lower Silesia, Poland. Well preserved original structural elements make this object an interesting subject of research consisting in the determination of changeable environmental influences that affect this type of structures. As it can be seen in Figure 1, presenting the truss structure in 1945, the II World War did not cause damage to the structure of the truss itself, although the sheathing suffered some serious destruction. The roof structure stands out above the local urban buildings, so it is exposed to potential influence of wind and sunlight. Measurement results discussed in the article are based on data collected in the course of ongoing long-term research of a predicted duration of 18 months. The collected data reflect the static work of the structure under the influence of such external factors as wind load and temperature changes. The analysis presented herein is based on data recorded during two selected days. Although this analysis is of a preliminary nature, the obtained results are interesting.

There are many different approaches to monitor the health of structures. The most common is to determine the deformation progression of important structural elements. Many of those approaches, as described in comprehensive review [4] of technology developments in monitoring large-scale bridges. The described technology ranges from fiber optic sensors to a global positioning systems (GPS). However, for many of the historical timber structures, using described technology is not feasible, therefore authors of this article present a different approach.

DESCRIPTION OF THE ANALYSED STRUCTURE

The trusses cover a gothic church building, characterized by a nave layout, with a main nave and two aisles (Fig. 2). The truss structure lays on four masonry elements: two external walls and two rows of columns supporting a masonry arch structure.

It should be noted that wall plates and sole-plates can be considered as the only immobile points of the truss structure. The span between sole-plate axes is 10.5 meters. Horizontal elements of the truss divide it into 8 tiers: the first four of a height of 3.2 m each, followed by three of a height of 2.90 m each and the last one 2.15 m high. The truss consists of two structurally different types of bearing structures (Fig. 3 and Fig. 5). The bottom part (up to the fourth tier) is a rigid frame, crowned with a double spandrel beam and stiffened lengthwise with double purlins at each end. The top part is a multi-collar beam structure with a longitudinal king post frame supported directly on the double spandrel beams. The king post and the posts of the lower frame structure are coplanar and constitute to a full members roof truss occurring in the spacing of 3.95 m. In between the full members trusses, there are three trusses of reduced members. Those are supported by the lengthwise stiffening elements (purlins) that are fixed and stiffened in the elements of full members girder.

The technical condition of the structural elements varies depending on the given element. Also, it can be estimated that approximately 15% of the elements were replaced or repaired during structure renovation. The repairs can be divided into the following categories: (1) replacement of the whole element, (2) partial replacement of the element (dominant group), (3) filling in gaps in compressed elements, (4) adding steel gusset plates on both sides of each rafter, (5) reinforcement of the cross-sections by adding additional side, top or bottom timber elements, particularly in nodes, and (6) supplementing the dowels.

Generally, it may be stated that, due to its age and state of repair of the structural elements as well as the clearances on the joints between these elements (in nodes) the truss is an object characterized by quite a large extent of free movement.

RESEARCH STATION

The application of draw-wire sensors for the measurement of displacements in a construction of such large size required to erect a steel structure constituting an immobile point of reference that would serve as a base for the installation of sensors. This structure consists of a steel framework bridge suspended between two central sole-plates (centers of 10.5 m apart), on which a light, aluminum tower was erected, additionally braced and stiffened to prevent it from moving. Due to the dense grid of horizontal structural elements in the timber truss and the necessity to keep the tower structure independent from the roofing structure, the height of the tower was limited to the level of the first obstacle.  As a result, the tower is 13 m high; it reaches the fourth tier and ends at 1m distance from the double spandrel beam of the framework structure. The additional height required in order to enable the equipment to reach the structural elements of the multi-collar beam structure was obtained by means of constructing two 5 m long parallel tubular masts. The masts were stiffened transversely in two directions with use of tubular angle struts. The immobility of the tower was confirmed by a series of test measurements. The tower structure was used only to install draw-wire sensors, while the recording equipment was installed on existing working bridges supported by the truss structure. This allowed to control the equipment without the need to climb the tower and stiffen it with additional struts each time, which would be necessary in such case and would lead to the decalibration of the measurement base of the sensors and thus to the discontinuation of conducted measurements.

Displacements of the structure were measured with use of two types of draw-wire sensors manufactured by Posital: LM0-AC005-3D2D-ZANARW and LM0-AC005-1D2A-ZACPAM. What is interesting is the fact that draw-wire sensors are mainly used as distance sensors for the determination of the position of objects with respect to each other, mainly in mechanics [11]. Thus, their scope of application is very specific and has nothing to do with the measurement of structure displacements. However, their construction and method of measurement make them a perfect device for capturing movements in large-size structures. In this case, the developed measurement method consisted in placing the sensor on the prepared, immobile tower (measurement base) and reaching the measurement points located on the structures with use of wires. The longest wire used is 12 meters long. The sensors were subjected to a series of preliminary tests whose aim was to determine the actual measurement parameters and thus to eliminate measurement errors. Detailed information about these issues and tests results was discussed in publication [13].

Each sensor was equipped with a specially designed bracket (Fig. 4) to facilitate it’s installation on the tower structure and to set an appropriate angle of the measurement direction. The construction of the bracket prevents the sensor from moving or rotating accidentally.

Wind speed and direction was registered with the use of a weather station DAVIS Vantage Vue with Davis Weatherlink Datalogger that allows for 36-hour data recording and enables connection with a computer for continuous data registration. The station was installed on the roof of an adjacent building, whose height corresponds to the height of tier zero of the church roof structure. The weather station has been installed approx. 50 m from the church roof, in the central part of the roof of the building, in order to eliminate wind turbulences that occur on the edges of the roof. The internal software of the station allow measurement of maximum and average wind velocity at one-minute intervals. For each of these values the direction of the wind is also recorded.

The remaining measurement instruments are standard test station equipment. One of the elements worth noting is the point-to-point temperature distribution measurement system that monitors the way in which the structure heats and cools down. The location of the temperature sensors is shown in Figure 5.

POSITIONS OF DRAW-WIRE SENSORS ON THE STRUCTURE

The position of the sensors was carefully considered and confirmed by a preliminary numerical analysis of the behavior of the structure in conditions of wind load and temperature changes. This enabled the identification of  points characterized by maximum displacements. The structural model was calculated and analyzed as a three-dimensional frame. The obtained results confirmed that the multi-collar beam structure was characterized by the highest deformability. The authors used the approach based on a symmetrical location of measurement points. A drawing presenting the positions of these points on the structure is shown in Figure 5. It was decided to measure displacement values along three directions (depending on the element type): horizontal, vertical and perpendicular to the analyzed element. A total of 11 measurement points recording the deflection of rafters (6 points), the horizontal work of the posts (4 points) and vertical work of the king post (1 point) were installed for each truss. Measurement points on the central (with reference to full members girders) girder with reduced members were placed in the same way in order to provide the possibility to conduct a comparative analysis of the work of both structural elements at identical geometric locations. Additionally, a draw-wire sensor of a constant measurement base of 10 m was installed on the test site, with the aim to compensate for the thermal work of wire elements.

The main axis of the church building is located in the East-West direction, therefore the exposure of each of the roof slopes (northern and southern one) to sunlight differs dramatically. Thus, in order to enable precise measurement of the response of the structure to uneven temperature changes, each of the eight measurement points located at the roof slopes is monitored individually with respect to local temperature changes. The test site is also equipped with an overall temperature sensor and relative air humidity sensor, placed in mid-height of the structure. These sensors provide information about average changes in temperature and humidity.

ANALYSIS OF THE RESPONSE OF STRUCTURAL ELEMENTS

The recording equipment allowed registration of all values measured by the sensors. Continuous online access to the equipment also enables to remotely track these values and to control data recording. Structure displacements are measured at a frequency of 1Hz. Air temperature changes are recorded once every 15 minutes. Recording and averaging of data concerning wind direction and intensity are conducted at one-minute intervals. Sample 24-hour recording of displacements of measurement points is presented in Figure 6. The following convention of measurement and designation of displacements was adopted for the purpose of the study: a change in the location of a measurement point in the external direction with respect to the roof structure is considered a positive change. Similarly, a negative change on the diagram of displacements refers to movements of structure towards the interior of the truss.

The observed changes in the displacements of structural points can be divided into two categories: (a) Changes lasting for several hours, in form of smooth long-period waves. Their main cause may be the changes in temperature resulting from exposure to sunlight; (b) Short-term changes, in form of smaller or larger peaks that occur at irregular intervals, most likely caused by the influence of wind.

The further section of the study presents the results of preliminary analyses of the displacements of selected measurement points located on the structure of the analyzed roof. The aim of the analysis is to demonstrate that such displacements are actually caused by environmental factors.

As the results of displacement measurements recorded by the equipment are obscured by noise interfering with the actual measurement signal, further analysis was based on signal that had been subject to a smoothing process with use of a moving median of a window width of 61 samples. The causes and methods of noise removal were discussed in detail in the publication [13]. Sample result of smoothening the measurement signal is presented in Figure 7.

INFLUENCE OF TEMPERATURE

The influence of temperature on displacements of the roof truss is presented basing on daily data of the 28.09.2014, which was generally still, and the maximum recorded momentary wind velocity equaled 1.8 m/s. The previous partial analyses of structure behavior show that displacements caused by wind velocity lower than the threshold value of 3 m/s are beyond the measurement range of the used measurement devices. Figure 8 shows a wind rose of the directions and velocities of average one-minute winds in the analyzed period, confirming the absence of wind that might significantly influence structural displacements.

The course of temperature changes in measurement points (their positions are shown in Fig. 5) located at the roof rafters are presented in the diagram in Figure 9. Temperature courses for southern slope sensors are marked in red, while those on the northern slope are marked in green. The observed daily amplitude of temperatures ranged from 7.7ºC for the sensor placed in mid-span of the bottom rafter on the northern side (RP2) to 14.9ºC for the sensor placed in mid-span of the upper rafter, on the southern side (LP6). The mean daily temperature amplitude was 12.0ºC.

Figure 10 shows the daily displacements of measurement points located on the rafters on the partial members girder, southern slope (LP) and northern slope (RP). It can be noted that the displacements of individual measurement points are nearly devoid of local peaks that might confirm the influence of the wind on the movement of the analyzed structure.

Figure 11 presents comparative diagrams of temperature and displacement changes for partial members girder at individual measurement points. Actual time courses of the displacements are marked with grey (cf. Fig. 10) green and red lines (cf. Fig. 9) mark temperature changes. A close, inversely proportionate relation between the displacements of measurement points and temperature changes can be observed. In order to increase the visibility of this relation, an additional diagram of displacements was plotted on the existing diagrams, in form of a horizontal mirror reflection of the original diagram of displacements of the given point (marked in black).

Correlation and determination coefficients were calculated for the presented displacement and temperature measurements. Their values are presented in Table 1. They confirm the existence of a significant, proportional relation between temperature and displacement, which is proven by correlation coefficients nearing one (without taking into account the negative or positive sign, indicating the direction of the measured displacement). High values of the determination coefficient confirm that the measured structure displacements are highly dependent on temperature changes. Moreover, it can be stated that at the observed temperature changes the structure works in the elastic state, returning to the original position corresponding to the given temperature value. In order to analyze the significance of correlation between displacements and temperature, an additional statistical test was conducted. For the assumed significance level α = 0.001 the critical statistical value t_0.001 equals 3.398. For all measurement points the obtained statistics t (see Tab. 1) is significantly higher than the critical value t_0.001. Thus, it can be stated that there is a 99.9% probability that the relation between the temperature and displacement of the analyzed structure points is statistically significant.

INFLUENCE OF WIND

In order to illustrate the differences in measurements depending on the presence (or absence) of wind, two analyses will be presented in this section of the study. The first refers to a situation in which no wind was recorded or the intensity of wind did not significantly influence structure displacements. The second one refers to a situation where the influence of wind was sufficient to cause displacements of measurement points on truss structure.

In the first case, analysis was based on data of the 28.09.2014 that were used in the previous analysis concerning the determination of temperature influence, although the time period was narrowed to eight hours, i.e. from 1:00 to 9:00. During this period, minimum and maximum temperature changes in individual measurement points amounted, respectively, to 1.9ºC and 2.3ºC (cf. Fig. 9). In the analyzed period, the maximum average minute wind velocity recorded by the weather station was 1.3 m/s and average – 0.44 m/s. Figure 12 shows comparative diagrams of time courses of displacements of measurement points located on the same tiers, on both sides of the partial members girder (green – northern slope, blue – southern slope) and average minute wind velocities (marked red in the diagram).

Significant influence of wind on structure displacements was presented based on the measurements from a seven-hour period (from 1:00 to 8:00 p.m.), recorded on the 19.06.2014. As it is shown in Figure 13, the dominant directions of average minute winds were south-west to north-west winds. The average velocity was 3.5 m/s. Figure 14 shows the wind rose of the directions and velocities of average one-minute winds in the analyzed period. These winds should be classified as very weak.

The visual analysis of the diagrams presented in Figure 12 confirms the hypothesis that the main factor influencing the displacements recorded in the analyzed period (1:00–9:00) is the influence of temperature. Wind influence is, in this case, either insignificant, beyond the measurement capacity of the equipment used or it simply does not exist, due to too low wind velocity.

Figure 15 shows the displacements of measurement points located on the rafters in the reduced members girder on the southern slope (LP) and northern slope (RP) and the average temperature changes in the analyzed period. It can be noted that the displacement diagrams are characterized by a significant number of leaps in their values and a high frequency of changes both in the amplitude and direction of the displacement. A specific dynamics which is characteristic for wind influence can be observed. Generally, the diagram is characterized by an increasing trend, typical for thermal influences analyzed earlier in this paper. Thus, it can be stated that the above diagram presents a superposition of thermal influences and the impact of wind.  In order to distinguish between temperature and wind influences, basic courses of displacements were smoothened with the use of a moving median of a window width of 121 samples, which has also been shown in Figure 15. It should be noted that the double increase in the window of the moving median, from 61 to 121, resulted in significant smoothening of the displacement course diagrams.

Correlation and determination coefficients were calculated for the displacement and temperature smoothened in the way described above. Their values are presented in Table 2. In this case the values also confirm the existence of a significant, proportional relation between temperature and displacement (correlation coefficients nearing -1) Moreover, the high values of the determination coefficient confirm that the smoothened displacement courses are highly dependent on temperature changes. Thus, one may claim that the relation between the temperature and the smoothened course of displacements of the analyzed structure points is statistically significant. Due to the above, time courses smoothened with use of a moving median of window width of 121 samples can be treated as a record of displacements of measurement points caused only by temperature changes.

Figure 16 presents diagrams showing the differences between the full, recorded displacements and calculated displacements resulting from temperature changes. In this case it may refer to displacements caused only by the influence of wind. The diagrams have been grouped in pairs for measurement points located on the same tiers, on both sides of the reduced members girder (top – southern slope, bottom – northern slope). The amplitudes of these displacements fall into the range between 0.6 mm and 1.0 mm. The last pair of diagrams shows changes in average minute wind velocity, which ranged from 0.4 m/s to 7.6 m/s in the analyzed period.

CONCLUSION

The study contains results of the analysis of displacements on one of the largest historical timber roof truss structures in Lower Silesia, Poland. Displacements of selected geometrical points of the roof truss recorded with the use of draw-wire sensors are a superposition of responses to two dominant influences: wind and temperature. The collected data allow to distinguish between the results of both types of influences even for very weak winds, of average minute velocities ranging from 2 m/s to 5 m/s. Analyses confirmed the existence of a close, statistically significant correlation between the recorded structure displacements and temperature changes.

The results of the analyses presented in this study confirm that draw-wire sensors may be successfully used to measure displacements of civil structures in situ with a sufficient accuracy. Furthermore, the quality and accuracy of the obtained and analyzed data confirms that draw-wire sensors can be used in structural health monitoring system.

REFERENCES

Accepted for print: 26.04.2016

Responses to this article, comments are invited and should be submitted within three months of the publication of the article. If accepted for publication, they will be published in the chapter headed 'Discussions' and hyperlinked to the article.