1 Introduction

Nowadays, it is generally agreed that existing buildings of heritage value constitute an integral part of the historic legacy of humanity to be preserved for many further generations to come. With modern societies targeting further the matter of sustainability, the conservation of such buildings appears to be a good alternative from the economic, environmental, social and cultural points of view. Even if this concept only came up in the last decades, a generalized degradation and abandonment state were already reported for the greatest part of the built heritage. Due to lack of maintenance, often leading to excessive moisture content exposure, various pathologies of a biological, physical or chemical nature may develop within timber structures, resulting in the reduction of their long-term mechanical performances in their service life [1]. Although structural disorders (e.g. instability, high deformation, accidental failure of elements or joints…) caused by a bad design at the initial construction stage are rarely met, they may still appear as a consequence of a significant loading change, over many intervention periods in the edifice history.

Prior to any decisive intervention in buildings of heritage value, it is crucial to develop a multidisciplinary method that combines several tasks and aspects from different fields of knowledge. With regard to the diagnosis and restoration of historic timber structures, many guidelines and methodological approaches can be found in the available literature [2,3,4,5,6]. Furthermore, two charters from ICOMOS [7, 8] also give some principles for the analysis, conservation and structural restoration of architectural heritage, with a focus on the preservation of historic timber structures. With a focus on timber roof structures, many case studies and state-of-the arts [9,10,11,12,13,14,15,16] introduce efficient methodologies of inspection and diagnosis, while promoting decisive interventions mostly driven by the conservation principles, with strengthening or repair possibilities.

Wood can be considered as a very important raw material throughout the ages, with a fundamental structural role in historic buildings around the world. In that context, the present work aims at assessing the conservation state of existing timber structures, and finally proposing appropriate prevention and intervention means to ensure good maintenance conditions in the near future. To this end, the Convent of Christ in Tomar (Portugal) was chosen, with the evaluation of three timber roof structures badly preserved over time. Thereby, a methodology of inspection and diagnosis was proposed and applied to these case studies in four main steps: (1) visual inspection; (2) non-destructive wood diagnosis; (3) safety structural evaluation; (4) prevention and intervention measures. Before going ahead with the assessment of timber roof structures, a few words should be said about the history of the building, namely the Convent of Christ, and the research methodology selected in the present work.

2 Materials and method

2.1 The building: Convent of Christ

Standing at the top of the hill, the Convent of Christ (Fig. 1) overhangs the historic city centre of Tomar, in Portugal, as a strong national symbol of “Reconquista Portuguesa”, and afterwards of the opening up towards other civilizations through overseas explorations. The whole historic building is the testimony to an eclectic architecture over a 500-year period, combining Romanesque, Gothic, Manueline, Renaissance, Mannerist and Baroque elements. The twelfth century Tomar Castle was built by D. Gualdim Pais, of which as first traces of the monumental complex the Oratory “Charola” and the walls surrounding the convent still remain to date. In the fourteenth century, the Order of Christ substituted the Order of Knights Templar, through progressively restraining the activities of friars to religion only. The next three centuries disclosed, step by step, the construction of many edifices which will give to the convent its heritage grandeur and appearance as we know today.

Fig. 1
figure 1

Source: A Terceira Dimensão, http://portugalfotografiaaerea.blogspot.com

Aerial view of the Convent of Christ, in Tomar, Portugal.

Full size image

Thereby, different elements of the monastic complex (Fig. 2) can quickly be stated in a chronological order: (1) centre—Romanesque Oratory “Charola” and the Gothic Capitulum Room; (2) east—two gothic cloisters; (3) south—Gothic Church in ruins and the gardens; (4) north, west and south—five renaissance styled cloisters forming with the Charola a huge Latin cross set of new dormitories; (5) north-east—infirmary and new pharmacy including the Knights’ Hall. Further information (not detailed in the present work) about the history and composition of the convent can be found in the following literature [17,18,19].

Fig. 2
figure 2

Second floor plan of the Convent of Christ in Tomar, with the location of timber roof structures under investigation (red areas)

Full size image

After the Portuguese Liberal Revolution, the Order of Christ was extinct in 1834 with the forced departure of the religious friars. From the early beginning of the twentieth century, some edifices were then occupied by either the Army or the Overseas Missions Seminar. In the 1980s, the Portuguese State reacquired the whole monumental complex, designated as Convent of Christ, for cultural and touristic activities that still run to date. It should be noted that the Convent of Christ was classified as a national monument in 1907, and only much later in 1983, registered in the World Heritage Site List of UNESCO. With the progressive abandonment of unoccupied convent parts in the 1970s and 1980s, many timber structures came up with pathologies, such as leaking water in the roof covering that triggered the onset and development of biological wood deteriorations over time.

In that context, three timber roof structures featuring major disorders have been investigated in the present work. They are, respectively, located in the dormitory Room 68 on the northern flank of the convent and in the Knights’ Hall at the north-eastern corner of the infirmary complex (Fig. 2). Through applying an inspection and diagnosis methodology to those case studies, the resulting analysis and proposal of interventions could be extended in the near future to some auxiliary rooms that include similar roof structures and pathologies in the Convent of Christ. Concomitantly, a wood dendrochronological survey has mostly been carried out in Room 68, while very recently a further geometrical assessment of the ceiling structure through laser scanning and drill resistance testing has been achieved by Cuartero et al. [20] in the Knights’ Hall.

2.2 Inspection and diagnosis methodology

Many methodological approaches for a mono- or multilevel characterization of timber structures in situ can be found in the literature [2,3,4,5,6]. Including a variety of actions and solutions, different methods may be feasible if they have been chosen wisely on a case by case basis. Regarding the assessment of timber roof structures in the Convent of Christ, a methodology of inspection and diagnosis (Fig. 3) has been proposed in three main steps: (1) preliminary evaluation; (2) structural analysis and detailed diagnosis; (3) results and future interventions.

Fig. 3
figure 3

Methodological steps required for a complete inspection and diagnosis in historic timber structures (modified from [5])

Full size image

  • The preliminary evaluation firstly consists in a desk survey, through gathering all handwritten and/or computer documents available (e.g. blueprints, reports, pictures…) dealing with the history of the building and timber structures investigated. Afterwards, a visual inspection takes place in situ to report the current state of timber structures in their service life, and if appropriate, their pathologies (e.g. structural disorders, biological wood deteriorations, serviceability defects…). A geometric survey is also performed in situ to identify and measure the structural elements and joints. Furthermore, it helps in defining better the main construction and/or modification stages of the structure in the building history;

  • The structural analysis and detailed diagnosis mainly aim at completing the information already collected in the first step. Before performing the structural analysis, the impact of the pathologies observed on the visual inspection on the wood mechanical properties has to be investigated. To this end, a wood diagnosis is carried out in situ through using non- and/or semi-destructive tests wisely selected on a case by case basis [21]. In detail, the timber structure is analysed on numerical software to detect the elements or joints featuring potential problems that could threaten the structural safety and integrity.

  • Based on the results obtained from the last two steps, preliminary and diagnosis reports are written as the assessment of the timber structure is in process. In addition to this, the third step aims at proposing a work plan of feasible interventions, with respect to the restoration principles stated in both charters from ICOMOS [7, 8], to solve current problems and prevent the onset and development of any further pathology. Thereby, the safety and integrity of the timber structures will be preserved in the future through enabling maintenance actions from time to time.

3 Visual inspection

3.1 Room 68

In the dormitory Room 68 located on the northern flank of the Convent of Christ, the roof structure and the ceiling planks (Fig. 4a) were selected as the first case study due to their bad state of conservation in the twentieth century [17, 18]. To enable the visual inspection and diagnosis in situ, it was decided to remove some ceiling planks that covered the rafters and joists of the roof structure (Fig. 4b). Also, only the northern and western parts of the structure were investigated, since they were highly subject to water leaks due to lack of maintenance of the roof covering. This made the onset and development of biological deteriorations easier over time on those areas.

Fig. 4
figure 4

Western rafters and northern joists made visible before (a) and after (b) removing the ceiling planks

Full size image

Based on the 3D modelling of the old roof structure illustrated in Fig. 5, different timber elements can be divided into two categories: (1) ceiling structure; (2) primary roof structure. The joists and edge beams belong to the ceiling structure, whereas the struts, purlins, rafters, collar ties and wall plates make up the primary structure which formerly counteracted the roof covering and outside loadings. It should be noted that all the timber elements are fixed to each other, side by side, with nails. The birdsmouth joint, which links the rafter end with the wall plate, also includes one or two nails to avoid any out-of-plane displacement of the generating truss constituted by the rafter, collar tie and strut of the rafter. The remaining purlins formerly bearded into three local points the rafters subject to bending due to the action of the roof loading.

Fig. 5
figure 5

3D modelling and horizontal blueprint of the old roof structure in the dormitory Room 68

Full size image

As shown in Fig. 6a, two different timber structures coexist underneath the roof of Room 68. The old structure under investigation had been built in the first half of the sixteenth century with the dormitory rooms, whereas the new one was recently set when restoring the whole roof covering in 2007 [18]. Since the new structure bears the roof and outside loads, the old one has been restrained in its initial structural role, being now subject to its own weight and those from the ceiling planks only. Also, the top part of the rafters was cut and removed from the old structure, because they were severely damaged due to biological agents. Nowadays, fungal decay at the rafter ends and insect wood deterioration on the external surfaces can still be observed (Fig. 6b), resulting in a significant reduction of the timber element cross sections. Indeed, many rafter ends, including birdsmouth joint and some parts of the wall plates are featured by a deterioration level so high that some of them (I, IV, V, VI and X) are nearly free of any support, leading to potential structural disorders.

Fig. 6
figure 6

Coexistence of old and new timber roof structures above the dormitory Room 68 (a). Rafters featuring insect wood deteriorations on their external surfaces and fungal decay at their ends (birdsmouth joint) with the wall plates (b)

Full size image

Old and new roof structures present different wood species. As per da Silva [18], oak was identified as the main wood species for the old structure, whereas the new one appeared to be made of pine wood. Furthermore, preliminary visual investigation concluded that the rafters, joist, edge beams, collar ties, purlins and struts could be made of chestnut (Castanea sativa Mill.), but of Portuguese oak (Quercus faginea Lam.) for the wall plates. Regarding ceiling planks, maritime pine (Pinus pinaster) is the most likely wood species.

This information was confirmed in the dendrochronological study performed by the researchers from the University of Coimbra, through sampling wood carrots within the old roof structure, mostly in the rafters. As another outcome from this study, average wood density of 613 kg/m3 can be assumed for all the timber elements made of Castanea sativa. As detailed in Table 1, this mean value is reliable when compared with the European standards (e.g. EN 350 [22], EN 1912 [23] and EN 338 [24]), although it seems higher than the value recommended by the Italian Standard UNI 11035-2 [25].

Table 1 Wood density \(\rho\) for the chestnut (Castanea sativa Mill.)
Full size table

3.2 Knights’ Hall

3.2.1 Octagonal dome

In the north-eastern corner of the Convent of Christ, the timber structure covering the Knight’s Hall was chosen as the second case study in the present research project. The “dome” stands for a eight-face pyramidal structure, whose top is featured by a void due to the geometrical assemblage of the ceiling planks (Fig. 7a). On the bottom part of the timber structure, decorative paintings classified as artwork under heritage protection can be observed on the ceiling planks (Fig. 7b). The octagonal dome and the ceiling paintings had been completed in the second half of seventeenth century [17, 18], in the same period as the convent infirmary and pharmacy. It should also be noted that some conservation and restoration works were performed on the structure between 1965 and 1970 [19].

Fig. 7
figure 7

Upper view of the octagonal dome (a). Paintings on the ceiling planks constituting the visible bottom part of the octagonal dome in the Knights’ Hall (b) from [20]

Full size image

Based on the visual inspection, each component of the octagonal dome has been modelled, as shown in Fig. 8. Thereby, it can be stated that: (1) eight pyramid edges constituting three beams each; (2) six panels of nine joists (faces I–II–III–V–VI–VII); (3) two panels of ten joists (faces IV–VIII); (4) 12 wall and 4 support beams make up the base of the structure. Again, the timber components are connected side by side with nails, while the nailed birdsmouth joint links the ceiling joists and pyramid edge beams to the wall and support beams. The octagonal dome was designed to only bear the decorative ceiling planks and the pendant light, exempting any structural role as regards to the roof covering. Since the pyramidal ceiling structure in the Knights’ Hall and the roof structure in the dormitory Room 68 share similar texture, colour and dating of wood, it can thus be inferred that the wood species is still chestnut (Castanea sativa). On the other hand, the decorative ceiling planks are probably made of maritime pine (Pinus pinaster). From the work of Cuartero et al. [20], some attempts in assessing a detailed geometry of the irregular cross section of timber elements, with non-visible faces and affected by biological wood deterioration, were carried out very recently, through a campaign of laser scanning and drilling resistance testing for the whole octagonal dome.

Fig. 8
figure 8

3D modelling and horizontal blueprint of the octagonal dome in the Knights’ Hall

Full size image

As structural disorder, the valley rafter from the northern roof under collapsing process tends to locally press the top of the octagonal dome (Fig. 9a), in the vicinity of three pyramidal edge beams between the faces V and VI, leading to a transfer of extra loading. On the other hand, slight and moderate insect wood deteriorations can be noticed on the external surfaces of the joists and edge beams. Furthermore, the joists (7–8–9) from face V have been indicated as highly degraded due to the same biological agent, resulting in a significant reduction of their cross sections (Fig. 9b). Also, those timber elements may occasionally be wet due to water dripping from the badly maintained roof covering. If those pathologies stated for the visual inspection still remain in the future, the structural integrity of the dome as well as the conservation of the classified paintings on the ceiling planks may be threatened in a long term, resulting in severe and irreversible consequences.

Fig. 9
figure 9

Beam of the northern roof structure pressing the top of the octagonal dome (a). Severe insect wood deterioration on the external surfaces of the joists from panel V (b)

Full size image

3.2.2 Carpentries set

A complex timber carpentries set, standing above the octagonal dome in the Knights’ Hall, was chosen as third case study. After assessing the geometry, position and orientation of each timber components in situ, the whole carpentries set was modelled successfully (Fig. 10). As a result, 50 timber elements were identified in total and sorted out into six groups: (1) 2 horizontal beams; (2) 12 purlins; (3) 4 ridges; (4) 15 struts; (5) 2 timber trusses; (6) 4 valley rafters. For each group, the timber components are connected side by side with nails. It should be noted that the purlins, ridges and horizontal beams are supported at their end on masonry pillars that appear to be settled and covered with lime mortar.

Fig. 10
figure 10

3D modelling and horizontal blueprint of the carpentries set above the octagonal dome in the Knights’ Hall. King-post truss (Truss A) and A-shaped truss (Truss B)

Full size image

Among this myriad of timber elements, two categories of truss stand out: (1) the king-post truss (Truss A); (2) the A-shaped truss (Truss B). The king-post truss (Fig. 11a) is made up of two rafters (1 and 5), one collar tie (22) in tension and one king-post (23) connected with four struts (24, 25, 26 and 27) that bear the respective valley rafters (37, 49, 43 and 30) from the roof covering. Furthermore, metallic U-shaped binding strips with bolts were used to strengthen the single step joints between the rafters and collar tie, and this technique restrained any out-of-plane displacement of the king-post end. On the other hand, the A-shaped truss (Fig. 11b) includes two rafters (17 and 18), one collar tie (19) in compression and two struts (20 and 21) that hold the respective purlins (45 and 47) from the roof covering.

Fig. 11
figure 11

Components of the king-post truss (a) and the A-shaped truss (b)

Full size image

Although the Knights’ Hall had been built with the infirmary and pharmacy in the second half of the seventeenth century [17, 18], the existing carpentries set appears to belong to the nineteenth or twentieth century. Indeed, the rendering of the timber components looks much more refined, suggesting that bench works in wood or modern manufacturing methods (e.g. sawing and machining) were used for that purpose. Even if there is no certified manuscript or report, the roof structure might have been built in 1970, during the major works for the repair and substitution of the whole roof in the infirmary and pharmacy, including the Knights’ Hall [19]. With the exception of ten struts (2, 4, 14, 15, 19, 20, 21, 28, 29 and 48) that are made of chestnut (Castanea sativa), eucalyptus (Eucalyptus globulus Labill.) is the main wood species observed for the whole carpentries set.

Apart from four timber elements (14, 15, 49, 50) that superficially deteriorated due to former insect attacks, no significant biological degradation has been noticed within the whole carpentries set. On the other hand, many structural disorders can be reported: (1) out-of-plane displacement of the king-post end (23) within the king-post truss (Fig. 12); (2) slightly twisted timber elements (1, 5, 6, 11, 22, 35, 45) over their whole length (Fig. 12); (3) tensile crack at the mid-span of the horizontal beam (11) under excessive bending (Fig. 13a); (4) collapse of the purlins (36, 38, 39, 41, 42, 44) due to the partial or complete failure of their respective masonry supports (Fig. 13b). All these pathologies previously stated constitute the proof that the northern part of the roof has already started to collapse since a while. As transient prevention mean, metallic struts were thus added accordingly, by holding on two purlins (35, 44) in the vicinity of the masonry supports.

Fig. 12
figure 12

Collar tie (22) and both rafters (1, 5) slightly twisted over their whole length. Out-of-plane displacement of the king-post end (23)

Full size image
Fig. 13
figure 13

Tensile crack of the horizontal beam (11) under excessive bending (a). Collapse of the purlin (42) due to the failure of the masonry support (b)

Full size image

4 Non-destructive wood diagnosis

Among the different non-destructive methods proposed to establish a better diagnosis of historic timber structures [21], ultrasonic pulse velocity, drilling resistance and impact penetration tests have been selected and performed in situ, with respect to the observations previously made from the visual inspection. Thereby, the non-destructive wood diagnosis mostly focuses on the roof structure in Room 68 and the octagonal dome in the Knights’ Hall, since they both feature significant wood deteriorations due to xylophagous insect and decay fungi. On the other hand, the carpentries set, reported as a sound and recent timber structure in accordance with the visual inspection, was disregarded from this step.

4.1 Ultrasonic pulse velocity test

Ultrasonic pulse velocity (UPV) test enables measuring the propagation time, and thus the propagation velocity of stress waves inside wood, between the transmitter and receiver probes [26,27,28]. Furthermore, the UPV method can be considered as a quality index of wood, since it can detect any superficial or internal defect (e.g. cracks, knots, biological wood deterioration, foreign body…). Indeed, lower velocities or longer propagation times disclose bad conditions of wood in terms of physical and mechanical properties. The propagation velocity of stress waves \(v\) (m/s) essentially depends on the density \(\rho\) (kg/m3), stiffness and thus the elastic modulus of wood \(E\) (N/mm2), which can be estimated by the simplified empirical equation (1) [28]. According to the same authors, the elastic modulus could then be correlated with the compressive and tensile strengths of wood parallel to the grain through empirical equations, featuring low and moderate coefficients of determination:

$$E = v^{2} \cdot \rho .$$
(1)

Due to heterogeneous contact conditions on external surfaces of the structural element (gaps, rough or rounded surface), it is sometimes challenging to ensure a good propagation of stress waves inside wood between both probes. For that reason, UPV testing, through using the equipment Pundit Lab® with two 54 kHz frequency transmission transducers, was only performed for the rafters located in Room 68. Conforming with the work of Feio et al. [28], two types of UPV tests were performed in situ to estimate the wood compressive elastic moduli parallel and perpendicular to the grain: (1) indirect method parallel to the grain (Fig. 14a); (2) direct method perpendicular to the grain (Fig. 14b). It should be noted that both methods require having access to at least two opposite faces of the investigated element. In all tests, a constant pressure was applied by means of a rubber to make the transmission of ultrasound inside wood easier between both probes.

Fig. 14
figure 14

Wave’s propagation with respect to the indirect method parallel to the grain (a) and the direct method perpendicular to the grain (b), through using the equipment Pundit Lab®

Full size image

As a result, elastic moduli parallel and perpendicular to the grain, noted as \(E_{{{\text{c}},0}}\) and \(E_{{{\text{c}},90}}\), respectively, were calculated by Eq. (1), taking into account the average wood density \(\rho\) = 613 kg/m3 given in Table 1 for chestnut (Castanea sativa). Besides, each part of the rafters assessed through UPV tests were sorted out into four biological wood deterioration groups (Fig. 15), with respect to both compressive elastic moduli \(E_{{{\text{c}},0}}\) and \(E_{{{\text{c}},90}}\): (1) sound; (2) slightly damaged; (3) moderately damaged; (4) severely damaged. The lower bound values regarding the “sound” group were inferred from the characteristic and mean values of both properties \(E_{{{\text{c}},0}}\) and \(E_{{{\text{c}},90}}\), with respect to the strength class D18 assigned to the chestnut (Castanea sativa) in accordance with the Standard EN 338 [24]. It has been shown that biological wood deteriorations, mainly due to insect attacks, are randomly distributed on the external surfaces of studied rafters. Nonetheless, three rafters (X, XI, XII) and most of the rafter ends in the vicinity of the decayed birdsmouth should draw some attention, since the related external surfaces have been reported as severely damaged. Furthermore, low values of \(E_{{{\text{c}},90}}\) recorded may also be explained by the presence of internal cracks in the element cross section.

Fig. 15
figure 15

Classification of external surfaces exposed to biological deteriorations, according to the wood compressive elastic modulus parallel (\(E_{{{\text{c}},0}}\)) and perpendicular (\(E_{{{\text{c}},90}}\)) to the grain, for the rafters in Room 68

Full size image

4.2 Drilling resistance test

The drilling resistance (DR) test consists of measuring wood drilling resistance as a function of penetration depth [28,29,30,31]. Since the diameter of the drilled hole in the element does not exceed 3 mm, the DR, regarded as a quasi-non-destructive test, has no impact on the wood mechanical performances or on the integrity of heritage structures assessed in situ (Fig. 16).

Fig. 16
figure 16

Resistograph® 3450 RINNTECH used for the drilling resistance tests

Full size image

Information given from DR test includes: (1) the wood drilling resistance; (2) the cross-section dimensions of the timber element studied; (3) the detection and location of inner defects or structural discontinuities (e.g. knots, cavities, cracks, deteriorated wood…) that could not be noticed in the visual inspection; (4) the annual wood growth rings; (5) a qualitative interpretation of wood density.

Whereas visual inspection often overestimates the amount of strength loss due to decay, DR testing may disclose significant drilling resistance in the reduced cross section [30]. In that case, preservation of the timber element should be promoted, instead of removing it permanently. To this end, the resistance measure (RM) [28] is calculated by Eq. (2), which takes into account the grey area under the drilling resistance profile (Fig. 17) and the drill penetration depth noted \(h\) in the element cross section.

Fig. 17
figure 17

Drilling resistance graph with respect to the drill penetration depth, including resistance measure (RM) values with and without defects

Full size image
$${\text{RM}} = \frac{{\mathop \smallint \nolimits_{0}^{h} {\text{area}}}}{h}.$$
(2)

From the literature [28, 29, 31], low and moderate correlations between RM, wood density and mechanical properties have been established up to now. Meanwhile, assigning the wood strength class from standards (e.g. EN 1912 [23] and EN 338 [24] to the elements investigated should not be based on qualitative RM values only. On the other hand, the wood moisture content should be assessed when performing DR testing, since it conditions the parameter RM as well as the wood mechanical performances.

Regarding the old roof structure (Room 68) and octagonal dome (Knights’ Hall), DR testing was performed for the rafters and joists, through using the Resistograph® 3450 RINNTECH (Fig. 16), in the vicinity of the birdsmouth joint. After plotting the drilling resistances on graph (Fig. 17), two different parameters RM were calculated by (2), one without defects and the other one with defects (e.g. internal cracks and biologically deteriorated wood). For both parameters, Table 2 provides the mean values, maximum, minimum, amount of performed DR tests, standard deviation (SD) and coefficient of variation (COV). Based on very similar data distributions with COV < 10%, it can be stated that the RM value with defects is, on average, 20% lower than the RM value without defects, resulting in a likely decrease of the wood mechanical performances as previously reported from UPV tests. As another outcome from DR testing, the thickness of the external surfaces featuring insect wood degradations should be taken as 10 mm on average, when estimating the residual element cross section.

Table 2 Summary of the average resistance measure (RM) from the drilling resistance test performed in Room 68 and Knights’ Hall
Full size table

4.3 Impact penetration test

Impact penetration (IP) test, also known as Pilodyn® method or dynamic pin penetration method, consists of applying an impact force through the release of a spring by a slender steel rod on the external wooden surfaces [28,29,30]. Being also considered as quasi-non-destructive, the IP test enables to measure the penetration depth of the rod, which is inversely proportional to the density and hardness surface of wood. On the other hand, the penetration depth increases with the wood moisture content, since the latter conditions the hardness and other mechanical performances of wood. Until a maximal penetration depth of 30–40 mm, this method can be used to detect the presence of superficial defects (e.g. cracks, knots, biological wood deterioration). Based on the penetration depth of the rod \(d_{{{\text{p}},{\text{rod}}}}\) (mm) obtained from IP tests, the related wood density \(\rho\) (g/cm3) can be estimated by the empirical Eq. (3) [29], with wood moisture content of 12%.

$$\rho = - 0.027102 \cdot d_{{{\text{p}},{\text{rod}}}} + 0.727987.$$
(3)

As regards the octagonal dome in the Knights’ Hall, IP testing was performed for the joists, using the dynamic indentor Pilodyn® 6J Forest (Fig. 18) in the vicinity of the birdsmouth joint. After taking measurements of the penetration depth, the wood density was calculated by (3), while ensuring a wood moisture content of 12% on the external wooden surfaces using the electrical resistance moisture meter Protimeter Surveymaster®. The mean values, maximum, minimum, amount of performed IP tests, standard deviation (SD) and coefficient of variation (COV) are given in Table 3 for three parameters: (1) penetration depth of the rod \(d_{{{\text{p}},{\text{rod}}}}\); (2) wood density \(\rho\); (3) external layer thickness of deteriorated wood \(t_{\text{D}}\).

Fig. 18
figure 18

Dynamic indentor Pilodyn® 6J Forest used for the impact penetration tests

Full size image
Table 3 Penetration depth of the rod \(d_{{{\text{p}},{\text{rod}}}}\), wood density \(\rho\) and external layer thickness of deteriorated wood \(t_{\text{D}}\), inferred from the impact penetration test in the Knights’ Hall
Full size table

The maximal density \(\rho\) = 619 kg/m3 that has been inferred from the minimal penetration depth \(d_{{{\text{p}},{\text{rod}}}}\) = 4 mm may belong to sound wood, within the range of density values detailed in Table 1 for chestnut (Castanea sativa). The external layer thickness of deteriorated wood \(t_{\text{D}}\) was then determined by reducing the measured penetration depths \(d_{{{\text{p}},{\text{rod}}}}\) by 4 mm, as a reference value standing for sound wood. As a result, the parameter \(t_{\text{D}}\) may vary between 0 and 11 mm, with a mean value of 5 mm. Since empirical equations from the literature [28, 29] have presented low coefficients of determination up to now, it has been decided not to estimate the wood mechanical properties based on the penetration depth.

5 Structural safety evaluation

After completing the preliminary inspection and non-destructive wood diagnosis, the safety evaluation of three timber structures, namely the old roof structure in the dormitory Room 68, the carpentries set and octagonal dome in the Knights’ Hall, was carried out on the Finite Element software RFEM 5.19.01 [32], by taking into account the impact of biological wood deteriorations and other disorders noticed in situ on their mechanical performances. To this end, the ultimate limit states (ULS) and serviceability limit states (SLS) were checked for each timber roof structure, in accordance with EN 1995-1-1 (Eurocode 5) [33]. From this standard, the coefficient of material \(\gamma_{\text{M}}\) = 1.3 and the modification factor for duration of load and moisture content kmod = 0.6 (assumption of permanent action and service class 2) were chosen to calculate the design values of wood strength properties, on the basis of the characteristic values provided by EN 338 [24] and UNI 11119 [34]. For each case study, the minimal design bending strength \(M_{\text{Rd}}\) (kNm) was estimated with respect to the lowest timber element cross section. Besides, both values delimiting the range of maximal deflection permitted to still ensure a good serviceability of the structure in situ, namely \(w_{L/300}\) = L/300 and \(w_{L/150}\) = L/150 (mm), were taken into account, for which the parameter L (mm) stands for the span of the timber element under bending on two supports.

5.1 Roof structure: Room 68

Within the old timber roof structure located in the dormitory Room 68, the numerical analysis focused on the rafters and ceiling joists in the northern and western parts suffering from a lack of maintenance over time. From the preliminary inspection and diagnosis, the element cross sections were determined, with the strength class D18 as mechanical properties attributed to sound chestnut (Castanea sativa), in accordance with EN 338 [24]. Furthermore, the roof structure belongs to the visual grade I (UNI 11119 [34]), for which three characteristic values of chestnut strength properties were considered: (1) compression strength parallel to the grain \(f_{c,o,k}\) = 11 MPa; (2) tensile strength parallel to the grain \(f_{t,o,k}\) = 11 MPa; (3) bending strength \(f_{m,k}\) = 12 MPa.

Nonetheless, the elastic moduli were modified in accordance, through partitioning each rafter, based on the results obtained from the ultrasonic pulse velocity test. Concerning the definition of the residual element cross section, a remaining capacity factor with a value between 0 and 0.5 could be assigned to the area attacked by a biological deteriorating agent [35]. To simplify the current model, it was decided to reduce the cross sections of the rafters and joists by an average thickness of 1 cm that stands for the external wooden layer featuring insect degradations with no remaining capacity factor, conforming with the outcomes obtained from the drilling resistance test.

Because the old structure does not have to bear anymore the roof covering loadings since the restoration works performed in 2007, only two weights should be considered: (1) weight of the roof structure (6.13 kN/m3) made of chestnut (Castanea sativa); (2) weight of the ceiling planks (0.2 kN/m2) made of maritime pine (Pinus pinaster). As illustrated in Fig. 19a and b, the internal forces (ULS) and local deformations (SLS) obtained after running the numerical analysis in RFEM present very low values, namely a maximal bending moment (M = 0.07 kNm) and maximal deflection (w = 2 mm) over 2000 mm span length, in comparison with the minimal bending strength (\(M_{\text{Rd}}\) = 0.4 kNm) and the inferior boundary value of the maximal deflection range (\(w_{L/300}\) = 6.7 mm). Although the whole roof structure has remained safe so far, several decayed birdsmouth joints linking the rafter end to the wall plate should be repaired to avoid any great disorder in the near future.

Fig. 19
figure 19

Internal bending moments (a) and local deflections (b) within the old roof structure in Room 68

Full size image

5.2 Carpentries set: Knights’ Hall

Regarding the carpentries set standing above the octagonal dome in the Knights’ Hall, severe structural disorders have been noticed, such as the tensile crack at the mid-span of the horizontal beam (11) under excessive bending and the failure support of the purlin (42) that both lead to a progressive collapse of the northern roof part. To figure out how these disorders developed over time, two case studies were tackled: (1) initial design configuration of the carpentries set without any disorder; (2) carpentries set including one purlin (42) free of any support. Featuring no biological degradation, the element cross sections have been defined with the strength classes D18 and D30, for the chestnut (Castanea sativa) and eucalyptus (Eucalyptus globulus) respectively, from EN 1912 [23] and EN 338 [24]. In addition to the own weight of the carpentries set (6.4 kN/m3), the permanent load related to the roof covering (0.8 kN/m2) has to be taken into account, encompassing the weight of the wooden planks and ceramic tiles.

For the case study I (Fig. 20a, c), the maximal value of internal bending moments, located at the mid-span of the horizontal beam (11), almost exceeds 50% of the bending strength of the timber element (\(M_{\text{Rd}}\) = 9.2 kNm), while the related deflections reach the inferior boundary value of maximal deflection range (\(w_{L/300}\) = 19.4 mm). Although the structural safety has been checked, the initial design of the carpentries set presents a weak stiffness for both horizontal beams (11) and (6).

Fig. 20
figure 20

Internal bending moments (a, b) and local deflections (c, d) of the carpentries set in Knights’ Hall, for both case studies I (a, c) and II (b, d)

Full size image

Regarding the case study II (Fig. 20b, d), higher values of internal bending moments (M = 7.45 kNm) have been recorded, as equivalent to 80% of the bending strength of the horizontal beam (11), resulting in a tensile failure under excessive bending and lateral torsional buckling. Besides, significant deflections higher than the superior boundary value of maximal deflection range \(w_{L/150}\) = 38.8 mm have been reported. It has been concluded that the deficient anchorage of the purlin (42) in the masonry pillar has progressively failed over time, leading to the collapse of the northern roof part observed nowadays within the carpentries set.

5.3 Octagonal dome: Knights’ Hall

Within the octagonal dome above the Knights’ Hall, low and moderate insect wood deteriorations have been noticed on the external surfaces for most of the joists and pyramidal-edge beams assessed. Likewise in the roof structure from the dormitory Room 68, wood mechanical properties have been defined for the structural elements constituting the dome, as belonging to strength class D18 and visual grade I for sound chestnut (Castanea sativa), in accordance with Standards EN 338 [24] and UNI 11119 [34]. Conforming to the outcomes obtained from the drilling resistance and impact penetration tests, the element cross sections, previously determined from the visual inspection, were reduced by an average thickness of 6 mm, through considering the external wooden layer degraded due to insect attacks with no remaining capacity factor [35].

Since the octagonal dome stands for a ceiling structure, three own weights have to be taken into account: (1) weight of the structural components (6.13 kN/m3) made of chestnut (Castanea sativa); (2) weight of the decorative ceiling planks (0.2 kN/m2), made of maritime pine (Pinus pinaster); (3) decorative iron pendant light (0.5 kN). With the collapse of the northern roof part, the rafter valley (43) from the carpentries set tends to press the top of the octagonal dome. To model this structural disorder, an extra permanent load (3.5 kN) was applied on the upper part of the pyramidal edge beams between the panels V and VI.

As a result, it was shown that the maximal value of internal bending moments (ULS), at the location point of the extra load that simulates the consequences of the northern roof part falling on the octagonal dome, does not exceed 50% of the minimal bending strength, namely \(M_{\text{Rd}}\) = 3 kNm (Fig. 21a). The latter term has been estimated based on the lowest cross section among all the ceiling joists, while the extreme values of the maximal deflection range have been calculated by considering a maximal joist length L = 3200 mm. On the other hand, significant deflections (SLS) measured for three joists (III-7, V-8, VIII-3), from the panels III, V and VIII, respectively (Fig. 21b), almost reached the superior boundary value of the maximal deflection range \(w_{L/150}\) = 21.3 mm. Although the structural safety of the octagonal dome has been checked, the singular deflections observed for those three joists should be reduced to avoid any damage of the decorative ceiling planks in the future.

Fig. 21
figure 21

Internal bending moments (a) and local deflections (b) of the octagonal dome in Knights’ Hall

Full size image

6 Prevention and intervention measures

In this chapter, several prevention and intervention measures have been proposed for the three timber roof structures investigated, by taking into account the outcomes from the visual inspection, non-destructive wood diagnosis and structural analysis. As regards the latter, the safety and integrity of structures have to be preserved and/or enhanced, by preventing the onset and development of severe pathologies in the future. Many examples of prevention and intervention measures to preserve long-term performances of timber structures in their service life can be found as a non-exhaustive list from the literature review of Verbist et al. [1].

6.1 Minor interventions

As a first minor intervention, cleaning the roof in the dormitory Room 68 and Knights’ Hall must be performed by removing dust, mud, wooden waste or rubbles accumulated near the masonry support, since their presence, to say nothing of the adequate range of temperature and humidity, may trigger and speed up the long-term degradation of timber structures due to biotic agents, such as wood-destroying insects and fungi. Also considered as a source of future biological infestation, the external layer of joists and rafters, which has been reported as degraded due to insect attacks on visual inspection, has to be removed as well, using wood carving chisel tools for instance. Afterwards, surface and/or inner treatments against this pest should be selected wisely and applied in situ on the remaining timber elements. Besides ensuring proper air ventilation, monitoring the indoor exposure conditions of timber roof structures (i.e. ambient temperature, relative humidity and wood moisture content) should be promoted to faster detect wood deteriorations and stop their development over time. In the dormitory Room 68, the totality of decayed ceiling planks under the old roof structure should be replaced by new ones made of the same wood species (i.e. Pinus pinaster) and treated with respect to the current design configuration. On the other hand, conservation works should be held from time to time to preserve the decorative ceilings as heritage values, underneath the octagonal dome, in the Knight’s Hall.

Since the repair and substitution works performed in 1970 in the Knights’ Hall, the 50-year-old roof covering has been eroded over time, for lack of maintenance, leading to service nuisances such as leaking water within the carpentries set and octagonal dome. Therefore, it is now crucial to promote a maintenance campaign of the roof covering, which will help in preventing water leaks in the long-term and, thus, the onset and development of biological wood deteriorations threatening the mechanical performances of timber roof structures in their service life. To this end, the roof covering has to be dismantled firstly, through sorting out the well-preserved ceramic tiles for reuse, from those that are damaged and have to be substituted by new ones. Once the structural disorders present within the carpentries set will be corrected and the timber structure returned to its initial design configuration, the waterproofing of the roof covering should be ensured, or else enhanced. To this end, an elastomeric waterproofing coating should be used underneath the ceramic tiles, on the covering of wooden planks. In addition to this, a mortar bed could be applied liberally to fill potential gaps between ceramic tiles in their final implementation. Lastly, a constant monitoring of all roofs from the Convent of Christ should be promoted in the near future to assess their conservation state over time.

6.2 Major interventions

Although the safety of the old timber roof structure has been checked in the dormitory Room 68, biological wood deteriorations, namely fungal decay, have been noticed for many birdsmouth joints at the rafter end linked with the wall plate. Since those connections are crucial in the stability of the whole structure, the rafters (I, IV, V, VI and X) featuring high advanced stage of fungal decay at their respective end should be either strengthened or repaired. Nonetheless, the entire removal and substitution of deteriorated timber elements from historic roof structures by new clear ones should be excluded for conservation reasons. Adapted from the interventions applied for the decayed timber beam end [36], different techniques can be sorted out into two main categories for the birdsmouth joint (Fig. 22):

  • Retrofitting through adding new partial or full-length timber element(s) made of the same wood species as the existing one, with or without fasteners/binding strips connectors (e.g. steel, FRP…).

  • Repair through settling prosthesis made of either the same wood species as the existing one or another material (e.g. epoxy resin mortar, concrete…), with glued-in plates or rods (e.g. steel, FRP…). Note that the connection of the prosthesis with the existing element can also be external (e.g. fasteners, screwed plates…) when the aesthetic value is not considered as an important intervention criterion.

Repair techniques are often seen as an onerous alternative that requires available passages large enough to easily reach the intervention region, namely when cutting the decayed part and substituting it by a prosthesis, which is sometimes very challenging to ensure within existing timber structures. Therefore, retrofitting techniques should be preferred when intervening in the dormitory Room 68. Regarding the octagonal dome in the Knights’ Hall, significant deflections have been reported from the structural safety evaluation for three ceiling joists (III-7, V-8, VIII-3) under bending, featuring very low cross sections. Again, their removal and substitution by new timber elements should be excluded for conservation reasons. Therefore, retrofitting techniques should be considered through adding new full-length elements, made of chestnut (Castanea sativa), connected alongside the existing ceiling joist with metallic fasteners and/or binding strips to reduce their current deflections. In that way, the conservation of decorative planks underneath the octagonal dome will be guaranteed.

Fig. 22
figure 22

Adapted from interventions on the decayed timber beam end [36]

Techniques to retrofit or repair decayed birdsmouth joints.

Full size image

For the carpentries set in the Knights’ Hall, the collapse of the northern roof part has been noticed in situ on visual inspection, and afterwards confirmed from the structural analysis through simulating a support failure. In that context, four main steps of major intervention measures should be promoted to permanently stabilize the carpentries set:

  • Holding the northern part of the roof structure back in its initial design configuration through using metallic struts or even outdoor crane, bearing the respective purlins and valley rafters, during the whole intervention period. Note that the roof covering (i.e. wooden planks and ceramic tiles) should be removed beforehand;

  • Strengthening the contact support between the purlins (35, 36, 38, 39, 41, 42, 44, 45) and the masonry walls with a metallic bracket anchored inside (Fig. 23). To ensure good anchorage conditions, metallic rods should be introduced inside the masonry wall, if adequate, and attached on their upper part to the bolts from the metallic bracket, which afterwards will be encased together in a concrete block;

    Fig. 23
    figure 23

    Strengthening of the purlin support with a bracket anchored in the masonry wall

    Full size image

  • Substituting the damaged horizontal beam (11) by a new element made of the same wood species (Eucalyptus globulus), but with a higher cross section to avoid any mid-span failure due to excessive bending underneath the contact area with the valley rafter. To enhance the compressive loads transfer, the contact surface between the horizontal beams (6, 11) and the respective valley rafters (37, 43) should be enlarged through adding some wooden wedges connected with metallic fasteners (e.g. nails, screws…) between both structural elements (Fig. 24);

    Fig. 24
    figure 24

    Substitution of the failed horizontal beam and enhancement of the compressive contact area with the valley rafter

    Full size image

  • Removing the metallic struts or crane that had been used to hold on the roof structure during the intervention. Lastly, putting back the roof coverings where they were before the intervention.

7 Conclusions

The present work has focused on the assessment of three historic timber roof structures, badly preserved over time, from the Convent of Christ in Tomar (Portugal) as a case study. From an adequate methodology of inspection and diagnosis, the roof structure in Room 68, the octagonal dome and the carpentries set in the Knights’ Hall were assessed successfully in four main steps: (1) visual inspection; (2) non-destructive wood diagnosis; (3) structural safety evaluation; (4) prevention and intervention measures.

For the visual inspection, the geometric survey was performed through identifying and measuring each structural component. As regards the pathologies noticed in situ, the old roof structure in Room 68 and the octagonal dome in the Knights’ Hall feature decayed birdsmouth joints and elements superficially deteriorated due to insect attacks. On the other hand, several structural disorders have been reported within the carpentries set in the Knights’ Hall, triggering the progressive collapse of the northern roof part over time.

Concerning the non-destructive wood diagnosis, the efficiency of three non-destructive tests was shown, namely the ultrasonic pulse velocity, the drilling resistance and the impact penetration tests, in the detection of wood defects within the three roof structures assessed in situ. Based on the results obtained from the ultrasonic pulse velocity test, the elastic modulus was estimated with respect to the wood density for the external surfaces of sound and biologically deteriorated elements. On the other hand, their residual cross sections were inferred through combining the drilling resistance graph with the penetration depth obtained from the other two non-destructive tests.

In the third step, the three timber roof structures were successfully modelled on the Finite Element software RFEM®, by taking into account the impact of pathologies noticed in situ on their mechanical performances. It was concluded that the progressive collapse of the northern roof part in the Knights’ Hall was triggered by a failure support, followed by a tensile crack of a beam under excessive bending, within the carpentries set. On the other hand, the safety of the other two roof structures investigated was checked, although some decayed birdsmouth joints should draw some particular attention in the future.

Lastly, some prevention and intervention measures have been proposed, based on the outcomes obtained from the previous steps, for the three timber roof structures investigated, to preserve and enhance their safety and integrity in the future in the Convent of Christ. Nonetheless, a strong collaboration between architects, engineers, historians and artisans is necessary to ensure economically viable, sustainable and, reversible if possible, solutions with respect to the restoration principles and present heritage values.