The roof of Notre-Dame de Paris. Analysing the structure and its destruction on April 15, 2019

Everyone knows the Cathedral Notre-Dame de Paris, at least from pictures. In terms of cultural history, it is of great importance. Its construction in the 12th and 13th century and all later changes are a central reference point of the development of the Gothic architecture. It is also in the focus of an entire nation and also has a high symbolic value far beyond France. This was evident in the many reactions to the catastrophic fire on April 15, 2019. However, some reactions also showed how little knowledge of architecture and construction technology there is in general. This video is an attempt to analyze and explain the structure that has been destroyed and to understand it in connection with the spreading of the fire. The ceilings here of the nave are rib walls that combine two façade segments to six part walls. Before installing these filigree components, however, the roof trusses had to be erected above it. A carpenter's masterpiece. It was necessary to deal with timbers with lengths up to 15 meters above the two parallel walls at a height of over 30 meters. The attic above the nave and the choir dates back to the time of the first roofing in the 12th and 13th centuries. To bridge the 13 meter wide nave in a stable way was difficult. Rafts were thereby statically less problematic than the tie beam that had to hold the medieval triangle roof truss together. At long length, these beams bend too much, so the first approaches for suspending these beams emerged. The tie beam is suspended from a beam that is often called king post, which in the simplest case distributes the load on the rafts at the ridge point. Because of the tie beam, there remain only vertical forces on the masonry. A secure connection of the king post is very important. This drawing shows a very massive king post in the apse of Notre-Dame de Paris. In the nave they worked, still a bit clumsy, with three hanging posts, of which the outer ones hung from the middle collar beams. Of course, it wasn't sufficient to rely on the rafters alone, so an additional triangle had to be inserted as well. These complete roof trusses were built only at every fifth place in the longitudinal direction of the roof. In between, much more simple trusses were arranged. Their forces were transferred to the complete roof trusses partly by the plates on top of the walls, partly by the longitudinal beams under the collar beams. Braces were also attached to these beams, providing lengthwise stability. Some of the lateral hanging posts were removed later, but there was one complete original roof truss in the nave directly before the crossing. A laser scan by Professor Andrew Tallon shows that the drawings from the 19th century were indeed correct. The artful but somewhat unsafe realization of the lateral hanging posts is interesting. By the way, you can see on the left one of the many holes that were drilled through the vault in the 19th century. Here, the said roof truss only recognizes two additional parts that were put in later. In the longitudinal direction of the roof, just north of the center, ran a bridge, or catwalk, which was supported by the tie beams. On the southern side of the center, almost all hanging posts were missing. Here you can see the braces for longitudinal stability and the joinery of the hanging posts. A very valuable construction research shows the entire structure in longitudinal section. On the right, there is the western gable with doors to the attic. On the left side is a part that has been altered in the 19th century, but we will discuss that later. Let's stay with the old parts and have a look on the roof of the dryer. The straight part was similar to that of the nave, but not identical. Particularly impressive was the design of the semicircle of the apse. Here, all collar beams were led very laboriously from the rafters to a central king post. This was supported backwards by three inclined beams in the longitudinal direction of the roof. This picture shows the level of the middle collar beams. The lower level, which can be seen here, was designed accordingly. At this place in the attic, there was a large hand-operated winch with an estimated ratio of about 1 to 40. It served for pulling up a rope through the walls of the apse, for example to raise flags or decorations. There were pulleys in the old attic, so that ropes could be lowered through the vault in other places as well. This one in the nave here, however, was probably related to this smaller winch at the bottom of the spire. This is the view from the apse back to the crossing. The bridges to the left and to the right led to small dormer in the roof. In the opposite direction, this image shows the pimpers one level up above the lower collar beams. One recognizes here, as well as in other pictures, the multiple repairs in the attic. The aforementioned construction research shows this detail in the cross section. Almost all joints between rafters and collar beams have been repaired. Now we look at the northern and southern transept. Their roof structure was completely replaced between 1844 and 1864 by the architect Eugène Villers-le-Duc. Here we are looking from Lowe's window in the northern transept gable to the south. In the foreground on the left, a railing is visible. There was an access from a spiral staircase. Another similar entrance was in the southern transept. Here we walk in the opposite direction to the Lowe's window. The new roof trusses of the 19th century were located here with a much greater spacing, as they had to correspond with the vaults. The longitudinal reinforcement was installed directly under the ridge. The roof trusses had an upper part, where a load from a center post was distributed sideways to two points, which were supported by long struts coming from below the thigh beams. Two posts were hanging from the same points, holding up the thigh beam. The thigh beam and the collar beams were doubled and took the struts and posts in the middle, connected by screws. This drawing shows the roof truss of the transept directly at the crossing. It had additional stiffening in the middle. The static advantage of these roof trusses over the medieval ones is clear. The forces from the struts were shifted to the top of the walls to these lower points in the corners of the vaults. In the special case of this truss, the posts were under compression, not tension, as additional struts ran from the base of the posts to the walls. This was possible here, since the resulting horizontal truss was easily absorbed by the longitudinal lathe or choir walls. A very nice insight into the space between the roof trusses and the vaults, for example, was possible at the southern transept, where wooden stairs led from the access door to the level of the beams. On the second truss, in the background, you can see an interesting chinery, which is probably no later repair, but was there from the beginning. The roof of the crossing was not only a roof truss, but also a foundation for the spire on the crossing. To ensure stability, even parts of the adjoining roof trusses were involved in the static concept of this foundation. This is a drawing of the vaults, in which the special situation of the crossing becomes clear. A four-part rib vault spanned an area of about 13 by 14 meters in a slightly irregular quadrangle. Therefore, only four support points were available for the roof truss in this area. Nevertheless, they had already dared to build a first crossing tower in the 13th century. This drawing of the substructure already shows all the important static elements needed to distribute the load over the four points. However, over the centuries, the tower suffered damage from wind loads, against which the structure was not ideally stiffened, and probably also rot, so it had to be demolished at the end of the 18th century. In the course of the restoration by Villers-le-Duc around 1860, a new second spire was built. Here is a floor plan of the substructure for this new tower. The crossing was spanned diagonally. A section across the diagonals shows the complicated structure of the corresponding roof truss, which merged seamlessly into the tower. Tapering upwards, the timbers reached a statically consistent continuation up to the top of the tower. In this animated scene, I have put together all the essential parts of the tower to show the complexity and elegance of the construction. I have left out some intermediate pieces and the braces against the torsion. This is a historical photo of the construction. It shows to which point the old roof structure was removed. Here finally the tower shortly after it was finished, with a cock made from copper on top of it. Let's go now inside, into the spire. Here is a look up the middle post. Only little space remained in the construction to build stairs. A narrow spiral staircase was located near the central post of the spire in northeastern direction. The staircase led up to the lower one of two platforms. In this wooden model, the platform can be guessed by the slightly less dense wood structure here. The sloping wooden beams formed an interesting double pyramid in the middle of the platform. The access to the smaller upper platform was through a round hatch with metal lid, as this interesting excerpt from a questionable video shows. Here again the model for a better understanding of the double pyramids in the middle of both platforms. Just below the tracery on the upper platform hung the three bells of the clock of Notre Dame. Here it becomes particularly clear how artfully all exposed parts were covered with lead. This work, which requires a lot of skills and craftsmanship, was executed by a company called Gaget-Gautier-Essy in Paris. However, this company became famous a little later with another work, the Statue of Liberty. About 250 tons of lead have been used for the spire, but this information isn't very reliable. Perhaps this number contains the other decorations for the roof as well. Here is a chimera in the tradition of pagan demon defense, as it was quite common in churches in the Middle Ages, similar to the numerous gargoyles. On the western end of the roof ridge, above the apse, stood this beautifully crafted cross. At the foot of the spire there was additional ornamentation that descended into the roof valleys on four sides and separated them into two valleys with an intermediate ridge. This ridge was practical for the proper drainage of water, as it avoided a valley situation directly above the main elements of the roof construction. At the same time, it was a consequence of the construction itself, as it was formed by those beams that were meant to stiffen the tower. On these beams there were some wooden pillars, on which stood twelve apostles and four mythical creatures. The topmost statue on the southeastern side was a bit special. She was the only one looking towards the tower. This statue of St. Thomas also represented the builder of the tower, Violet Leduc. Let's go back to the spire itself, this time its constructive basis. This initially consisted of two triple oak beams diagonally across the crossing. In this cross a beam quadrangle was inserted. It was supported by sixteen beams from the four corners of the crossing. This picture shows the triple beam at the northeastern corner. The year 1863 can be seen written on the timber on the left in this picture. Also interesting was the scarf joint in the upper layer. The next image for comparison shows the corresponding situation on the northwest corner. This time the scarf joint is in the middle position. On the left you can see the important stiffening beam, here embedded in the triple beam, which ran all the way from the vault to the lower tower platform. This image from the southwestern corner shows even better the system of the members under the triple beam. It can also be seen that the stone arch of the nave vault at the crossing is doubled with an additional stone layer. Here in the counterpoint in video as well as a panoramic view from the middle of the crossing to the east, north and west. Let's take another look at the middle post of the spire. There was a plate which bore the year 1859. Then Violet Leduc is mentioned as architect as well as the carpenters company Bellu. The timbers were designed by the carpenter Henri Georges. Obviously these were exactly the same people who had recently built the crossing spire of the Cathedral of Orleans. Below that was a symbol of the Compagnons, the guild of carpenters. A level above, some levers for wire pulls to strike the bells in the crossing tower and a heavy white, both an indication of a mechanical clock in the attic. Indeed, there was one in the glazed box in the crossing close to the southern transept. The clock was connected via long steel rods with four clock faces enormous on the transepts. The clock alone would be enough to tell for a whole video. It was manufactured in 1864 by the company Collin, formerly Wagner, in the Rue Montmartre. It was wound up once a day with 254 crank turns by hand. The poles for driving the hands lay protected in long wooden boxes, which were covered on one side with sheet metal. These wooden boxes ran through the entire transept roof, apart from the prominent roofing. Notre Dame, however, contains other carpentry, the timber frames in the west towers. These were built in 1850 in one piece, continuously from about the level of the eaves of the Maeve to just below the pyramidal roofs. Slightly tapering upwards, they are like separate inner towers and have a weatherproof covering with openings for the sound. Here we look at the lower part of the construction in the south tower. Farther up, in the upper floor, with the big arches, there are the bells, the big historical Emmanuel and nine more from the year 2012. In 2019, a large steel scaffold was erected around the crossing to restore this area. The scaffold stood on four scaffolding towers, which had been led up from the corners between Maeve and the transept. Already in 2018 and in January and February 2019, the attic had been extensively treated against rot and fungi with special chemical substances. In the course of the external scaffolding construction, scaffolding material was also extensively installed in the attic. You can see that all workers wore protective clothes and masks, probably because of the recently applied chemicals. Many temporary platforms were built by installing long aluminum traverses at different heights. There was also a temporary electrification with a distributor at the foot of the crossing tower. The attic had previously been without any electrical installations for fire safety reasons. The historic catwalks were covered with plastic material and secured with plastic nets to prevent material and tools from falling onto the walls. A first smoke detector alarm at 6.20 pm was classified as a false alarm, because no fire was found. It was not until the second alarm that a fire in the attic was discovered, but then it had evidently already spread very far in the crossing. Starting point of the first externally visible smoke were the roof valleys at the crossing tower. The first visible flames flared a few minutes later. The roof lead in the area of the roof valleys had now melted at the foot of the spire. With slightly greater intensity on the southeast side, as far as that can be judged. The roof trusses in the crossing were apparently already in full blaze. The fire began probably in the area of the clock and had a much lower dynamics at first. It is not easy to imagine the transition from a small fire to the full blaze of a rather flame-retardant oak structure. But from the point when the first flames came out, the further dynamics of the fire are completely understandable. Ventilation of the roof structure from below was provided through the many vents over the walls. After the melting of the lead and the burning of the roof cladding, the crossing provided the effect of a chimney, so that a very hot fire could develop just below the crossing tower, where the highest density of wooden construction was. At the same time, the fire spread out under roof ridges at the crossing. At first, the fire crept eastwards, under the medieval roof of the choir. The timber density in the transept was a little bit lower immediately below the ridges, so that their roof trusses did not light up so distinctly from the timbers near the ridge. Rather, the entire roof cladding ignited over the whole surface by the increasing fire heat from the inside. Here you can see the southwestern dormer. The clock came to a stop at 6.30. If this is the actual time, it is 10 minutes after the first and 13 minutes before the second fire alarm. That, too, could be a hint for a breakout of the fire near the clock. The strength of the fire is also indicated by the fact, that the light wind of 3 meters per second from east-southeast did not significantly influence the direction of the spread of the fire. The roof of the choir even burned down to the apse against the wind, while about half of the nave roof was still standing. From the roof of the choir, the half-cone over the apse collapsed at last. While in the straight roof the rafters often toppled individually, the massive construction over the apse fell as one part in a northwestern direction onto the vaults. When this part tipped over, yellowish-white flames with white smoke were observed, probably caused by the lead oxide from the lead plates. The crust on the roof ridge struck north on the masonry, where it stuck hanging over. A short time later, around 7.56, the spire tipped to the northwest and broke apart a few meters above the upper balcony. The upper part of the tower fell right into the nave. The lower part of the tower then plunged further south, approximately in the direction of the diagonal of the crossing. It got stuck in the scaffolding, which it fortunately met at a relatively stable location, not so the upper part. Fortunately, the top struck the northern wall of the nave, which somewhat eased the impact on the vault. And by the way, that saved the copper cock, which broke off and fell on the side roof. But this part of the tower hit the vaults of the main nave. The impact on the vault took place in the area of a belt arch on its north side, as well as on the adjacent vault, in which a hole of about 10 square meters was created immediately. The impact on the belt arch must have knocked out several stones immediately, after which the entire arch fell down. Both adjacent vault fields collapsed almost completely. The upper part of the crossing tower fell into the nave, together with about 100 tons of stone. The wood continued burning, but this fire soon could be put out by the firefighters. A robot was used, because it was to be expected that more vault parts would collapse. Perhaps the most spectacular recordings of the entire fire show how the fire on the roof can be seen through the hole in the vault, while through the round vault opening in the middle of the crossing, repeatedly embers are falling down. Another very small hole is already visible in the vaults of the crossing. It was perhaps caused by the lengthwise impact of a timber, the clock weight or a bell. The burning material in the nave was interspersed with scaffolding material from the roof, but not from the outer scaffolding, which stood up. With the following burning of the roof up to the west gable, a new situation was created for the firefighters. While the spreading of the fire previously had been unstoppable, now there was a need and at the same time an opportunity to prevent at least the fire in the belfries. This succeeded, albeit only partially. The north tower lit up at 9 p.m. in the lower area, exactly in the direction of the wind. However, the firefighters were able to extinguish this fire before it spread to the entire belfry. At about 10 p.m. there was a slow decrease in the heat on the vaults. At this time, several jet tubes from worklifts were aimed at the fire from all sides, and it was clear that it would become a race against time, if one wanted to preserve the vaults from overheating and collapsing. This would have jeopardized the stability of the entire structure, since the walls in the area of the upper level would have been in great danger of tipping, especially in the case of an asymmetrical collapse of the vaults. A domino effect would have been possible. It was not until 11 p.m. that the fire was under control and that the construction was not in danger anymore. A drone picture taken at about 10 p.m. shows the distribution of the fire. Some dark places are the hole in the nave vault and some remains of the roof, for example one half-moon-shaped part on the trier vault. An additional fire was probably caused by falling parts on the side aileroofs. Luckily, they have a different construction with stone slab covering. Incidentally, the firefighting attacks also took place through the doors in the west gable of the nave. When looking from the southern tower, it is noticeable how strong the fire still burns in the area of the crossing, even though it started there. But of course, there was also the biggest amount of wood. So it is understandable that during the night the vault of the crossing no longer withstood and collapsed, probably due to failure of the mortar joints in the apex. In the morning, the extent of damage was obvious. The entire vault of the crossing is missing. The stone blocks from the ribs can be seen in the fire debris. The modern altar depicting the evangelists is damaged, but still standing. Through the hole in the vault you can see the remnants of the trifle beam, which still protrude from the corners to the middle. Another vault hole was created in a field of the northern transept. It is unclear why the vault gave way at this point. But perhaps there was a slight pre-damage by settlement cracks, or it was hit at an unfavorable angle by some timbers. This can also be considered as a reason for another small hole in the trial. Here is a view from the former attic access in the southern transept. The view corresponds exactly to that shown in the first section of this film. A few questions cannot be clarified with the present images and can be answered only later with the cleanup. For example, where is the whole lead? Is it lying in big lumps on the vault? Or where are the bells of the spire? Are there any remnants of the clock? Or why did a small roof area not burn away near the staircase of the southern transept? In summary, one can say that the fire has destroyed a very important and historically significant part of Notre-Dame de Paris and caused considerable damage to the masonry. It had barely escaped and, with much luck, an even greater disaster that could have meant a partial or complete collapse in the event of an even longer fire or unfavorable fall of the spire. The determination of France to repair the structure should not hide the fact that the fire has lost something irretrievable. It is to be hoped that the reconstruction of the roof in the sense of monument preservation will be seen as a repair of the entire structure, which will not be significantly changed by the repair. This means that while there is no need to rebuild the roof truss in all its details exactly as it was, it should, however, conform to the original in form and function, including the crossing tower. The statues around the crossing tower, taken off one week before the fire and thus preserved, underline this. Where should Violet le Duc look if not to his tower?