Fire Science Show

223 - Heat-induced delamination in CLT with Antonela Čolić

• Wojciech Węgrzyński

Share episode

In this episode of the Fire Science Show we invite dr. Antonela Čolić from the OFR Consultants, to break down the performance of adhesives used in CLT in fire, what differences between the glues are observable at the microscale  and how they show up in real structure fires.

We compare common polyurethane adhesives: one that softens near 200–220 C and one that resists softening, crosslinks, and ultimately chars. Through thermogravimetric and calorimetric testing, we map pivotal transitions like glass transition and softening. Then we scale up. With small shear-lap coupons and meter-long cantilevers under controlled heat flux, we see how mechanical load amplifies normal strains at the bond line—especially in cross-laminated elements where grain orientation concentrates stress. The result is a clear picture of when heat-induced delamination begins, how it differs from char fall-off, and why heat flux often dominates the story.

Moisture emerges as a powerful, often overlooked driver. Using neutron imaging, we visualize vapor moving toward and across the bond line, slowing as it crosses the interface. That temporary moisture retention can make an adhesive appear to “fail at a lower temperature,” not from chemistry alone but from local pore pressure and hydration dynamics. We translate these findings into actionable guidance: specify adhesives that char rather than soften, control lamella thickness, consider parallel lamellas to preserve capacity after a ply loss, and model realistic heat flux and shear demands instead of relying on a single critical temperature.

If you design or review mass timber, this conversation gives you the tools to ask better questions: Which adhesive? What heat flux history? How much shear at the bond line? And how will moisture in use and during fire shift the thresholds you’re counting on? 

Interested in further reading? Got your back.


----
The Fire Science Show is produced by the Fire Science Media in collaboration with OFR Consultants. Thank you to the podcast sponsor for their continuous support towards our mission.

Wojciech Wegrzynski:

Hello everybody, welcome to the Fire Science Show. The science of construction, civil engineering, is a very interesting one. My mentor at the Building Research Institute, Professor Czonesky, used to say that we often do focus on our research on discovering why something works, not necessarily inventing new things, and this is true for the construction because you know 100 years ago they didn't have finite element method, they did not have your codes, but yet they were building buildings, many of which are still standing till this day. And discovering why that is came later. And in this case it it kind of continues. I remember a few years ago we were been doing with OFR a very big experimental project on timber slabs which had different adhesive used in them. And indeed in those experiments you could observe a big difference, big change in performance of the of the ceiling with the different type of adhesive use. And and that was quite a surprise to me. I was not I have not expected such an obvious measurable outcome of this experiment, yet it happened. Now I must say I have not understood why exactly that happened, why exactly there was that big difference, and it appears it was quite an interesting research project to understand why. And as you can imagine, today in my podcast I have someone who's been working on the answer of why this measurable difference was observed, and that is Dr. Antonela Čolić from the OFR. Antonela has just finished her PhD on mass timber and uh the and all the interesting things that happen at the glue line in the cross-laminated timber. So in this podcast episode we go very deep into the glue line and we try to answer uh why adhesives work and at what conditions they fail. We go into thermal, mechanical, moisture, all the interesting things that that happen into the timber, all powered through the science, all based on scientific research, multiple experiments across the scales. I think it's a very interesting episode, I'll bite you have to be warned, it's a little bit more technically difficult because we really go deep into the the glues and uh and technology, but well, if I have one chance to do an episode about at his in the fire science show, let it be a good one. So let's spin the intro and jump into the episode. Welcome to the Firescience Show. My name Wojciech Wegrzynski, and I will be your host. This episode is brought to you in partnership with OFR Consultants, the UK's leading independent fire engineering consultancy. With a multi-award-winning team and offices across the country, OFR are experts in fire engineering committed to delivering pre-eminent expertise to protect people, property, and the planet. Applications for OFR's 2026 graduate program are now open. If you're ready to launch your career with a supportive forward-thinking team, visit OFRconsultants.com to apply. You will join a worldless organization recognized for its supportive culture and global expertise. Start your journey with OFR and help shape the future of fire engineering. Hello, everybody. I'm joined today by Antonela Čolić from the OFR Consultants.

Antonela Čolić:

Hello, Antonela. Hey, hi, Wojek.

Wojciech Wegrzynski:

I'm very happy to have you in the podcast and I'm thankful that you've agreed to discuss your very recent PhD. Congratulations.

Antonela Čolić:

Thank you. Thank you. Thanks.

Wojciech Wegrzynski:

And uh I really want to talk about the lamination today in a relationship with the CLT and structural timber in general. And I'm kind of stressed because I don't, I'm not even sure if I'm supposed to use the word lamination in this interview. So am I am I allowed to use delamination in here?

Antonela Čolić:

Uh yeah, we were gonna call it lamination, but it's heat-induced delamination.

Wojciech Wegrzynski:

Okay, what's the Kvet?

Antonela Čolić:

So the delamination is the one that we actually use in the ambient conditions, and when we test products for the detachment of the bond line due to the environmental conditions, which is drying and heating in normal, I mean like everyday drying and heating. And then heat-induced one is due to the presence of external heating sources, fire. So those are the elevated temperatures. But we can call it for the simplification today, the lamination. I think it's just really important to make the distinction at the beginning.

Wojciech Wegrzynski:

Yeah, thank you very much because there were some extremely complicated terms uh we started to use to this, uh, which definitely have their merit, and we will definitely go into that merit, but there's also uh power in simplicity, and if everyone calls it lamination, it's lamination for me.

Antonela Čolić:

I do, I do, I do, I do. Yeah, yeah.

Wojciech Wegrzynski:

Yeah, so it's a very interesting phenomenon. Um, can you give me a little bit of background how you've ended up doing this as your uh PhD at the University of Edinburgh?

Antonela Čolić:

Yeah, yeah, yeah. I mean, uh to be honest, like it's it's it's a cliche, but like when I was a kid as well, I was always impressed by glue. And how can you add a substrate that is completely different, the two things, and then just glue it together? And I was like, always thought like that's never gonna work. And yeah, I always thought like you have to take your nail and nail your stuff into this. It's like it's I I always believe more in mechanical fixing, but yeah, glues for some reason they do work, and then um when COVID happened, I worked on this small literature review with uh with Luke. I worked from Croatia, I'm originally Croatian, and I kind of got into the topic then, and then I that was during my IMFSE period, and then I got to do master thesis with Luke. So after that, Luke, I'm I'm really glad. Uh he asked me whether I want to do a PhD with him, and then we ended up doing a PhD as well together, and it's such an expensive topic that actually we started as one of the future work topics from Felix Wiesner's work. So when he graduated, he graduated in 2018, and that's when I started with my IMFC. So it was just a continuation of of work.

Wojciech Wegrzynski:

Fantastic. Well, MFSC that explains a lot. Actually, I missed that part in the bio, but here comes the missing puzzle. Great, yeah, great. And I also had a Felix in the podcast ages ago, eons ago, like 200 episodes ago, we've talked about moisture and know that this is also an important part of your considerations. But uh let's let's get into that. First, perhaps let's try and settle the difference between the char fall-off and the heat-induced lamination. Because I assume those are two different processes, two different manifestations.

Antonela Čolić:

No, no, no, that's very important. Clarification, and thanks for asking. So the difference is that the heat-induced lamination is the detachment of the pieces of lamella which haven't completely char. So that means that the char layer hasn't progressed to the bond line yet, meaning that it happened at the temperatures which are lower than the typical 300 degrees isotherm that we take for the paralysis of wood. So that means that you have the reduction of the composite action even before the char has progressed to the bond line. Uh, char fall-off is just the char falling off. That can happen even in solid timber. It can be just small pieces of char that are detached in the matrix of the timber itself. So the char fall-off in the glute or CLT or any laminated product, that can happen before the char has reached the bond line, just the char falling off, or when it reaches the bond line, the whole charce falling off or progress the bond line, any any char detachment. And the difference is A, when it occurs, and B how it affects structural mechanics. So if you have heat-induced lamination, you still do have some mechanical properties preserved of your bond line at the point when you're losing it. For the char fall-off, once once the char progresses, you've lost all of your structural capacity. Char has no load-bearing property whatsoever. So that means that with a heat-induced delamination, you're progressively losing your cross-section, which is really important for residual load-bearing capacity of your structure.

Wojciech Wegrzynski:

Well, with char you lose the protective insulative layer, so there's also like consequences of losing that. But indeed, it already had no uh structural load-bearing capacity of any sort. Um, I think we maybe, maybe should have stepped back a once before to even define the bone line, because like we're we're we're gonna like discuss some very high-level things in a minute. Let's define why bone line is important and how does it mechanically affect the properties of your cross-laminated timber at a large scale? Is it only to keep the lamellas in place or does it have a bigger role?

Antonela Čolić:

Well, the bone line itself is this interface between timber and adhesive. So it's not interface, it's the phase. It's it's kind of inter intermixed because it adhesive penetrates timber differently in different regions. So there is no this like clear line where you can say, oh, it's just the adhesive failure. So when you when you have the failure at the bond line and you have the two pieces of wood in your hands, you will see that on both sides you have some splinters of wood. That means that there is this like weird phase in which you had this detachment, meaning that it's very important how your adhesive actually interacts with timber and that it's not only the adhesive, that it's also the timber that you choose for bonding of your element. So that's the bond line definition when I ref refer to it as such.

Wojciech Wegrzynski:

And uh does it play any role in mechanical response of the timber or just it must in that case?

Antonela Čolić:

Yeah, of course. Yeah. So the way we design our structure is you you have this composite action, and that means that the element, when it's bonded, it behaves as one unit. So it behaves essentially as solid timber. So there is no distinction across the cross section. And if there is a distinction that is made between the glue laminated timber and cross-laminated timber, where in the crosswise-oriented lamella, they have reduced elastic modulus. But in in the glue lam, everything goes in the one direction, all over the lamellas. So the the elastic modulus is the same across the cross section. But with the bond line, we never, as a structural engineer, I never had to take that into account that the bond line can have, depending on the adhesive choice, different elastic modulus. You just take it as the same across the whole timber. So in ambient conditions, you say that it plays no role, which adhesive you choose. You test the element as a whole. But then in the fire conditions, this was already found by Emberly back in the days, I would say five, six years ago, and Jose Torero, bond line, the failure mode in the element, in the structural element changes. So if it tests it in the ambient conditions, the failure is in is in timber. Once you start heating the element, the failure mode changes and it goes to the bond line.

Wojciech Wegrzynski:

So if you put force on your CLT slab in ambient, it will most likely break like a solid timber would.

Antonela Čolić:

Yeah, it yeah, it depends on which product you you test. Which engineered wood product you test, yes.

Wojciech Wegrzynski:

I don't know how many large CLT structures I've burned, but now I realize I've never crushed one without fire. So a non-fire failure is extremely you know uh exotic to me. I've never seen it fail without fire. It's probably funny to say, but yeah, that that that that's the case. But I've seen a lot of char fall-offs and I a lot of uh the lamination. So from that perspective, I'm I'm safe to carry on this discussion. Um maybe let's cover the adhesives then. What types of adhesives are used in this uh in this type of constructions? Do they have anything special to them? And basically, how how how does that world look like?

Antonela Čolić:

Yeah, yeah. So like back in the 50s, we had only glue-laminated timber, which are which which we use usually for beams and columns, and those are the line elements that we use. And we used to bond them with usually melamine urea formaldehyde, MUF, and that worked really well for like 40 years. But then in the late 1990s, the CLT was patented, and CLT had to use a different type of adhesive. And the reason for that is that MUF it cures with the application of heat and the pressure. And you can imagine that once you have the larger elements as CLT, you have this big amount of energy that you would need to cure such element, which is opposing to our sustainability idea and the reasons why we are using timber in the first place. So our the the new adhesive we came up with is one component polyurethane, and that adhesive it cures with the presence of water, which is naturally within timber. So it takes the moisture from the timber itself, and then you apply the pressure. So that would go good. You have also reduced formaldehyde emission, which was another restriction. And then we used the one component polyuretine for for years, but then in the early 2000s, around 2010, we realized that this adhesive doesn't perform that well in fire. The reason we figured that out is because we started building more complex and larger mass timber buildings and using more and more CLT. So then the research ramped up and we tested more and more products, and we realized that we need to figure out how to deal with this issue of heat-induced delamination, which was now a new phenomena occurring in CLT structures. So around 2020, we came up with this new adhesive, new one component polyuretin adhesive, commonly known as HBX. And we came up with a new testing method which proved that this adhesive works. So everyone now implemented that adhesive, and we were like, okay, fine, we are good. But then I spent four years of my PhD trying to figure out what are the conditions under which actually this adhesive works and why is that said.

Wojciech Wegrzynski:

Fantastic.

Antonela Čolić:

Sorry, that was like a long, widened answer. A really simple question.

Wojciech Wegrzynski:

Well, no, no, that that that's that's a good historical like uh like you would expect from someone who just did their PhD on something. Like you probably know more about those adhesives than uh the most people in the industry, which which is again expected. Thank you, thank you um for that. So now let's maybe try how the failure in that bond line may occur. So so uh what's the buildup to the uh to the failure and then how how the the failure itself happens?

Antonela Čolić:

So the reasons for failure can be either in intrinsic or extrinsic. So intrinsic ones are what is the type of the adhesive you are using, and what is the chemical composition of the adhesive you're using, what is the type of the wood you're using, and what is the morphology between the two. So that's everything that's happening, happening on the front of like what is the choice you're making. And then you have the extrinsic ones, and those are the external conditions that are applied on your bone line. So you have three things that are happening at the same time. First one is the thermal penetration. So you have the thermal wave passing through and exposing your adhesive and timber naturally to some form of thermal degradation. Then you have also the mechanical stresses that are naturally induced due to the load that you applied on your structural element. Then you have the thermomechanical effect, which is shrinkage and swelling, which that is induced now. And then you have the moisture movement, which is a really important one. Due to the presence of pressure gradients and the thermal gradient, you have the movement of this moisture from the heating source towards the bond line. So your bond line is at one point, it finds itself in this complex thermohydromechanical state due to the all of this effect, and it's and it's exposed to the complex set of stresses.

Wojciech Wegrzynski:

And let's start with the intrinsic ones. I'm really interested in the uh compatibility between the glue and the timber. So, what can go wrong in there? I I the most of CLT I've ever seen was uh spruce.

Antonela Čolić:

Yeah.

Wojciech Wegrzynski:

But in in my laboratory, there's a researcher, Dr. Pavel Sullich, and he's done a lot of research on locally sourced timber of different types, because you've also mentioned that we need uh CLT. Uh you said it between the lines, but we need CLT for sustainability reasons, and I fully uh fully agree with that, though I'm not sure if importing it through the continent from one factory in a very remote location is if that is the essence of sustainability. I think the future is like more locally produced CLT.

Antonela Čolić:

I completely agree, yeah.

Wojciech Wegrzynski:

Yeah, and I think the companies who are who are doing this also agree because I see the factories popping all over the place. Uh that's that's great news as well. But anyway, uh Dr. Sulik is looking into uh locally sourced timber, different types of timber in in Poland. Like I'm not sure if he was testing mechanical response, because that's perhaps where he would observe some differences. But in terms of their fire response, it was very interesting to see different. But he was making his CLT on his own, like you know, literally uh gluing uh different timbers together and pressing them. Anyway, what can go wrong between the adhesive and timber? Where is the risk of incompatibility between them?

Antonela Čolić:

So, first is the choice whether you go with a softwood or the hardwood. So the thickness of the selves of naturally between the two are very different. And that will then uh influence the infiltration of the adhesive in timber itself. And you can have this effect that we call starved bond line. So if you have in hardwood, it can occur that like your your your adhesive just penetrates so deeply into the hardwood that your bond line itself is starved, like if you don't you don't have anything anymore because it just like travels so deeply inside. And that's when you also have the gaps in the bond line as well. And then when it comes to the the the softwood itself, uh we tested Norway spruce and radiata pine. And the two we we found the differences between the two just due to the presence of different latewood and early woodwood ratios. So radiata pine was uh I did those experiments back in uh Australia, and radiata pine there, those are the trees that grow very fast. So the presence of the latewood, uh which is usually the during the winter times, is is is really low. And the early wood, which is spring and summer, that's that's really high. But the latewood is the one that is responsible for your mechanical response of your timber. And it's really, really cool. But like I think that this morphology aspect of it all needs to be further explored and it needs to, you know, like timber has its own natural variability, and you just need to know how to account for it. You you can't go into so many details to see, like, oh, is my bond line gonna be really close to the late wood or early wood and in which ratio? But I think we we kind of need to figure out how to account, or it would be at least fun to do research to to try and account for that natural variability of timber and and how close to the bond line this latewood and early wood is, because it does define the the response of your bond line.

Wojciech Wegrzynski:

Yeah, that's what I wanted to ask, actually. I wonder if the the type of timber is the defining thing, because you you may have it like stored outdoors versus stored indoors, like different moistures of the timber plank when it's glued. You may have it like uh a very anisotropic piece of timber, like the the like even within one tree, you will have a diversity of different structures in in the in the single planks. I also understand that CLT technology was meant to kind of get rid of this issue by you know just having a lot of different planks together and eventually all the things even it itself out. Are those like conditions at which the timber was stored before building in the CLT uh also playing a role in the bond line?

Antonela Čolić:

Uh definitely, yeah, yeah, yeah. No, no, for sure. And it's it's how was a timber stored before manufacturing? How is it stored? How is it pressed during much manufacturing? What is the dispense of the of the adhesive on the assembly line? What is the then the storage after the manufacturing? And then what are the using conditions in which like your your your timber is going to behave differently in the kitchen and in the living room, in in depending on where you are, what are your conditions of use? All of that affects what is the moisture content in your timber, and yeah, naturally, like when the moisture starts traveling, what is the pressure that affects the bond line.

Wojciech Wegrzynski:

So this kind of narrows the technology into factories where it would be very well controlled, I presume.

Antonela Čolić:

Yeah, we do believe that. I think Danny and Mike from OFR they did some research on whether the different supplier in Europe, if you choose different supplier, whether that would affect the consequential chart fall off. Is that what they were studying? And they didn't find much difference between the three different suppliers. Uh, but you know, that was the only study I know of that that that did that.

Wojciech Wegrzynski:

Well, large large-scale experiments with CLT are quite an expensive game, I would say. And you know that I know that very well.

Antonela Čolić:

Just to add, there is no study to my knowledge that compares different species across the world. So the three suppliers we use Danny and Mike used in their study, those are all from Europe. So it's highly likely that those are all spruits or like same-ish softwood. So at least if we are referring to that study, we can say, okay, we are safe in Europe, but you can't compare the standards and the methodologies across the globe.

Wojciech Wegrzynski:

Let me ask you a question. I it it's like curveball, so uh you may say you don't want to answer it, but can you actually glue hardwood to softwood and it's gonna work, or is no one knows?

Antonela Čolić:

I would not personally recommend.

Wojciech Wegrzynski:

Okay.

Antonela Čolić:

Why? I've not not. It's just completely different microstructure.

Wojciech Wegrzynski:

Okay.

Antonela Čolić:

So so the cells are very different around the bone line. Uh your adhesive is gonna interact differently. The mechanical interlocking, which is how your adhesive penetrates the cell itself, is gonna be different. The infiltration through the cell is gonna be different. So, you know, we already have a lot of, I mean, a lot of, I know don't want to make it uh existential, but there are already issues if you bought bond softwood to stuff to it. So we need to figure that out a little bit better before we progress to the combination of softwood and hardwood.

Wojciech Wegrzynski:

And uh last one, are the surfaces prepared in any way for that? Like, do you do any any time I I don't know English words for processing timber. Is there any processing that's done to the surface before adhesive is applied?

Antonela Čolić:

I believe it has to be sanded, but I uh yeah, I think that that's the extent of manology. I do believe that some some people depending on adhesive, primers can be used that first applied on on timber, but like the specific mechanical change of the surface of timber, I'm not sure.

Wojciech Wegrzynski:

Okay, let's move to extrasy. So you said it's thermal penetration, mechanical, including thermal mechanical response and moisture. Let's take them one by one. So, how does thermal penetration affect the bond line leading to its failure?

Antonela Čolić:

So the way we studied this is we started from micro scale and then we went to the intermediate scale. So in the micro scale, we only studied how the thermal penetration changes the adherent. Adherent, when I say I mean adhesive and timber. And then I'll just give a quick overview of like the complete scope, and then we can dissect what happened at each extrinsic stage. So that was the the micro scale. Then we went to the small scale where we had such a thin element, we did a little shear laps where which allowed us to take out of the equation the moisture movement, and we could then study only thermomechanical response. Then we went even larger, where I only had the heating of the timber and the moisture movement. And we did some experiments where we exposed those little um timber to neutron flow in the neutron beam. And then the last aspect was the intermediate scale, where we did thermo hydromechanicals, and those are where you have the presence of all three moisture movement and thermal gradient and mechanical stresses. So let's now go from the start. Thermal penetration. The way we studied that was through two different microscale tests. Those are thermogravimetric analysis, where you heat something and it loses mass, and as it loses mass, you can see what are the thermal degradation uh stages.

Wojciech Wegrzynski:

Did you do TGA in uh nitrogen or in oxygen?

Antonela Čolić:

We did both. We did both in nitrogen and oxygen, yeah.

Wojciech Wegrzynski:

Okay.

Antonela Čolić:

And the second one is differential scanning calorimetry. Uh, and that one is essentially at a heat flow, you get the response which can be exothermic or endothermic. Uh, and you it's telling you what is the molecular movement in your polymer chain when you're observing your adherent. So we in those two microscale methods, we tested separately adhesive films that we produce out of one component polyuretane and muf. But I'll stick, I'll I'll take MUF today out of the equation because it would just complicate the story. So I'll just refer to the two one component polyuretane.

Wojciech Wegrzynski:

Okay.

Antonela Čolić:

So we did the testing of the adhesive films, and then we also took the sawdust from the Norris spruce and from Radiatopine, and then we tested each of them separately to see what is the thermal response. So what happened is when we compare the behavior between the two adhesives, we could see in TGA tests that in the range from 200 to 220, for the HBS, you had this little plateau uh when you check your DTG curve. And that plateau usually means that the material is experiencing some sort of softening. But we couldn't confirm it because it's such a small temperature range, only from from 200 to 220. So then we tested it in DSC, and then what we observed in that same range for the HBS is that you have a little drop, and it's an endothermic drop, and that confirmed to us that the material has experienced some softening. So those two just those two simple micro-scale methods were really indicative of the behavior, and it gave us the answer why HBS usually fails around between 200 to 240 degrees. It was really clear that it's due to this transition, softening transition. But then before I even go to that softening, the DSC method also allowed us to see what is the glass transition temperature. And this occurred for both adhesives, HBS and HBX. And uh before you your material reaches the glass transition, it's kind of in a vibrant glassy state. And once it reaches the glass transition, your polymer change chains, they start to move against each other.

Wojciech Wegrzynski:

Sorry, is this done on the glue itself or glue on the timber?

Antonela Čolić:

No, no, this is all now just the adherent. Just the adherent. Just the adhesive film. And you could see the uh so glass transition temperature, once you reach it, the polymer change, they start to move against each other. And depending on the chemical composition of the adhesive, they can undergo additional crosslinking. So what we saw is that for the HBX adhesive, once it reaches this glass transition temperature, it does experience some crosslinking because the stiffness of the element it just which we observed in later stages in different experience uh experiments, it just plateaus. Whereas for HBS, there is nothing, like it just keeps the viscous behavior just continuous and the polymer chains they keep like moving against each other. So that for the cases HBS leads then to a softening temperatures, but for HBX it doesn't. HBX just chars, whereas HBS softens.

Wojciech Wegrzynski:

Okay.

Antonela Čolić:

This is kind of a small explanation of these like thermal degradation stages. There is a paper we published on that uh in the International Journal for Adheses, and I think that's quite useful to better understand actually.

Wojciech Wegrzynski:

Yeah, that's the exact paper uh that is in front of my eyes. Um, from a practical perspective, uh, as a fire engineer, if I do a thermal analysis of heat penetrating my timber element, and I just want to have a ballpark number to say, okay, it's most likely going to fail at this temperature, taking the lower bound of the range of softening, which you indicated at around 200 ish degrees, is that a reasonable approximate? How would an engineer apply that exquisite piece of knowledge you you just gained into practice?

Antonela Čolić:

Well, that that is really important because the what actually my whole piece of work showed is that just Doing the micro scale methods, just this thermal analysis, it's not sufficient to tell you at which temperatures, what are your what is your performance criteria, what is your critical bond line temperature. That is not sufficient. The reason for that is because you have these influences of the structural stresses and moisture movement, which we later observe on the larger scale. But if someone is asking me what is the temperature that I should use as a critical temperature in my models, for example, where I'm doing uh estimating what is the heat release rate curve, then I need to account for some additional fuel load that is that is added. If you really want to be super conservative, you can use the glass transition temperatures for both HBS and HBX. For HBS, that's 160. For HBX, that's 190. But that that's really super, super conservative. If you want to lose it a little bit, you could go with softening temperature for HBS.

Wojciech Wegrzynski:

That's the reason I I mean it's sometimes okay to be super conservative, but sometimes being super conservative is a fancy word for saying I'm wrong. And uh if you want to have a really detailed image of your the structural response of the element, perhaps you're being overly conservative with that. And as you said, there are other factors to consider which uh we will go in a second, but 180 versus 190 versus let's say 220 sounds not like a big difference, but in the in the realm of progression of thermal front into your structural element in a real fire, that can be difference of let's say 190 could be at a peak and 220 is never reached. That that's a possibility in a fire for sure, or it could be a 15-minute difference between these fronts uh propagate, it all depends on the fire. So so it it actually, while it's a little difference in temperature, timber is an excellent insulating material. So that's quite a difference, actually.

Antonela Čolić:

I do agree. Yeah, it has really good thermal conductivity.

Wojciech Wegrzynski:

Let's try to do mechanical stresses uh and and how that affects the lamination. First, uh a cheeky question. Do you need mechanical load for lamination to happen?

Antonela Čolić:

That's such a good question.

Wojciech Wegrzynski:

It's not mine, it's right.

Antonela Čolić:

No, it's it's it's really good because what happened actually is uh this is this is the study we presented at Interflan. So bone limb performance was usually tested through large-scale experiments, but those larger scale experiments which prove that the new adhesive, HBX adhesive work, are more often than not loaded with really low load, structural load. So then I was wondering, what if this adhesive performs only up to the certain stage, after a certain structural load? So in our larger scale experiments, what we did is we induced higher load and we figured out that once you induce really high shear stresses, the lamination occurs in both HBS and HBX. Whereas at small scale experiments that we did, and I say small scale are the ones that are just 30 millimeters, like they're really small shear lab samples where you have no presence of any moisture transfer. When we did those experiments, those were stressed at only 6% of shear strength, little veneers glued. And HBX performed well. It didn't delaminate at all. Like you didn't even have the bond line failure at 300 degrees, and if it failed, it failed outside of the bond line in timber itself. So whereas HBS continuously failed at those often temperatures that I was talking about, and that it would just disintegrate in the bond line uh phase. So those were the small-scale experiments where I had no presence of any moisture movement, and at six percent we saw no, like we really saw good performance of HBX. But then later, on larger scale, in I call it larger scale, but actually it's intermediate scale, we did two sets of experiments. One was the shear lap samples, which were half meter, around half meter, and then the other ones were the cantilever beams, which were 1.2 meters. And uh, you could see in the shear lap samples, we stretched it at 20% of shear strength, and all of them delaminated. No matter the heat flux, we applied 50, which was 50 kilowatts per meter square with a radiant panel that was representative of flaming combustion. We also had 25 smoldering combustion, representative of, and in both cases, in all cases, independent, regardless whether you use S or X, it all the laminates. And then we said, okay, 20% of shear stress, the strength that's really high. We did the analysis of what is I run the survey through the cost Helen action among the practitioners to see what is the usual law that we apply in our structures. And actually, the the shear, the the maximum shear that we we could see that was used in practice was ranging somewhere around 6%. But the rolling shear, which is another different type of failure that it can have in CLT, that one was going up to 13%. So what we decided is okay, we're gonna use these values now in our last set of experiments, the cantilever beam, to see how the increase of the structure load affects heat in use delamination. And what we realized is that the heat flux is the main driving parameter. So once you have 50 kilowaspermeter square, all of them delaminate, independent of the load that you put. But when you have on the 25 kiloes per meter square, the increase of the load does matter. So we tested it at 6% of shear strength and 12% of shear strength. And you could see that for HBX, once you increase the load, it really does define what is your behavior after you experience the heat-induced delamination.

Wojciech Wegrzynski:

One thing that that's interesting to me related to mechanical response and the lamination. So before the delamination, the plank of timber is still load-bearing, right? It still has some because uh two 200, how how much uh of load-bearing capacity of timber you would lose around 200, like more than half of it?

Antonela Čolić:

Yeah, it would be more than half at 60 degrees. You're already, I don't know by hand, but I think like you're you have less, but 40, if I remember correctly. Like that's that's like really, really low. Uh, but then you kind of stay on that. And I think that yeah, it at 200, you're still on like 20, 30, something like that. Don't take me my word.

Wojciech Wegrzynski:

Don't quote don't don't quote us in. I don't want to end up in jail. So I assume that uh I'm being this hundred percent, uh, 300 degrees is like zero, so there there must be some residual load bearing uh left before your glue is exposed uh to temperatures at which uh the bond line would fail. It must be quite interesting from the mechanical like the load-bearing partways, because you suddenly lose a part of your structure that was load-bearing, so you suddenly quite significantly reduce the cross-section of your element. And because it's a cross-laminated timber, the next plank is in a different direction. So it is not gonna necessarily behave the same as the one that just fell. So the logic when we were doing you know those large-scale experiments with OFR for many real-world projects, we were and we always done them uh loaded. Uh, we were like, okay, if we have five layers of timber in the CLD, we literally have the three load-bearing layers. That was a rough approximation. Uh and basically when we got into the third one, you know, when the third one became exposed, which means it starts drastically reducing its load-bearing capacity, we were really stressed because that meant we only have one plank at ambient at which all the load is. And this is usually the point at which the structures start making sounds. I really do not like my structures making sounds, like I like them to be silent, not moving, not deflecting, and definitely not making sounds. So perhaps this is outside of your scope, but where are you looking into how the the static uh load distribution changes when the planks fall? That doesn't could be an interesting study.

Antonela Čolić:

No, no, for sure. Yeah, I mean the stress distribute distribution in your element completely changes, and it's very different for CLT and uh GLULAN. So, yes, what we did is in in uh both of our intermediate scale experiments is is we we tracked the the strain in the bond line itself, and we did that by using the digital image image correlation. Um, not sure how much you know about it, but essentially you you you speckle the surface of your element in little dots, and you have your camera which is recording it, and you're just tracking how those dots move.

Wojciech Wegrzynski:

I I I love the technique, but it never works in fire lab because of all the smoke and dusting.

Antonela Čolić:

Smoke, exactly, yes. But we kind of tried to isolate it and we did get some results, so that was really, really nice, I guess. Uh and and and what we noticed uh is that even though the elastic stress that you impose on your element, which is due to the imposed load, even though your imposed load is very, very low, what happens is that when when you start heating your element on your bone line, due to the shrinkage and swelling, you have the normal normal strain which is acting on your bone line. And that's just due to the heating of your element. And that strain increases magnificently, much higher than the strain that is caused only due to the presence of the structural load.

Wojciech Wegrzynski:

Can you expand on thermo mechanical response to shrinking and expansion? Um, will there be a difference in shrinking or expanding at the bone line versus the virgin timber that's next to it? Does this cause additional stresses?

Antonela Čolić:

Yeah, yeah, yeah. So when you check the strain profile across your cross-section, it's just like the peaks are immense in the bond line itself, and then it relaxes in the middle of the plank, and then it again peaks in the bond line itself, meaning that the bond line is doing its work, it's it's it's taking a lot of strain. Yeah.

Wojciech Wegrzynski:

And is this also like something you would connect with the ability to withstand or promote the delamination?

Antonela Čolić:

Or yeah, I mean, definitely, and it's really promoted in the so it depends how stiff your adhesive naturally is. HBX is stiffer than HBS. We saw that in our ambient test, the displacement that we we got from the HBX was smaller than the one that we got from uh HBS. Um, and then naturally later in our experiments, there was a difference between the two and how they take the the strain and the bond line. The strain that was experienced by HBS was higher than the one that was experienced in HBX. And then the cool thing about the last part of our study is the in CLT, you have the big contribution of the fact that the lamellas are cross-oriented. So you you you naturally in timber have a different shrinkage in in tangential and radial direction. And then if you also cross-oriented, that then it's even more promoted because you have also the different orientation of your drains of timber. Whereas in GLULAM, everything flows in the one direction, so you wouldn't expect that to influence your adhesive that much. But then if we we when we compared in the glue lamp, because in the last set of experiments we tested both CLT and GLULAM to see what is the contribution of just the adhesive, because in glue in GLULAM you take out this problem of timber and differential shrinkage of the timber itself, right? So in the GLULAM, also see that the HBS has higher strain than HBX. So it is the adhesive difference as well as the contribution of uh of timber. I I hope that came somewhat clear. It's really hard to like explain strain in words.

Wojciech Wegrzynski:

People have people have been warned that we're gonna talk about high-level stuff on the online. So yeah, you signed for that. It was your choice to listen to this podcast, it's your fault. Let's talk about the moisture movement. I've covered that with Felix, but I would uh love a refresh on that.

Antonela Čolić:

Oh, that's that's so cool. I love it, and it's so non-discovered, and I know know so little about it. And and our experiments just show how little we know.

Wojciech Wegrzynski:

It's so hard to measure.

Antonela Čolić:

Yeah, so what we did is we did the experiments in um in France. We have a big uh uh nuclear reactor, and the way this method works is essentially you you have your element, you heat it from one side, and your moisture starts to move. Simultaneously, you point out the neutron beam through your sample, and it's kind of like doing x-rays for humans, but it's instead of checking the bones, you're checking the position of the hydrogen because these neutrons they are interact with the hydrogen in your element, and what you get on the back side is this image of what is the position of your hydrogen at every single moment of your heating. And simultaneously you're rotating your sample, and as you're rotating it, you get this 3D image of what is the position of the moisture throughout the whole cylinder that we tested. And what we saw is that as the moisture progresses towards the bond line, but once it passes the bond line, the the front movement is bilinear. So it has one increased rate up to the bond line, and then it reduces once it passes the bond line. The two reasons for that are first it's further away from the heating source, so the pressure gradients are naturally different. And the second is the presence of the bond line. So bond line does like the movement of moisture is not the same in the laminated timber and the solid timber. And then we wanted to see what is the difference between HBX and HBS. Is there a difference in the the bond line retention around the adhesive itself? And what happened is that actually both of them do have the retention of the moisture front at some point, but HBX keeps that moisture for the longer period. So then later, when we did our large scale, intermediate scale experiments, we saw that it looked like HBX fails at lower temperatures, but actually those were not lower temperatures due to thermal penetration, it was due to this retention of the moisture. So that's why it's so hard to say that the bond line temperature at failure was this and that, because it's that the system behaves really, really differently at every single scale.

Wojciech Wegrzynski:

Right. I'm just amazed, like you get to do you get to do experiments in nuclear facilities with Timber. That's so cool.

Antonela Čolić:

Yeah, yeah.

Wojciech Wegrzynski:

But we had to keep it at like 270 degrees, yeah, because fire and but the the the butt I assume the samples had to be fairly small size, so we were not able to load them either.

Antonela Čolić:

There are methods to load them, but that then you're like entering so many complexities, and we just wanted to dissect this hydrothermal response. So we didn't want to introduce load as well, but there is a way to introduce, they have instron that you can use whilst doing this screening as well. But yeah, I mean we we we opted not to do so, and the the the size was 20 times 50 millimeters, so it was just a cylinder, like it's it's really fine, yeah.

Wojciech Wegrzynski:

So you we have these three components the the thermal penetration mechanical stress moisture. Like how how do you now make out of these three variables one telling you uh how bad it is?

Antonela Čolić:

Like that well, that that's a good question. So so the larger scales are telling you what the response of your material could be at specific conditions, and then the smaller scale experiments they tell you why is that happening. So so the reason why the HBX had lower failure temperatures is not because it's weaker or whatever, it's because there is the moisture that is being retained stopped at the bond line. And you can only see that if you do different types of experiments, and now we know why that happens. Or when we see that that the HBS fails at 200 degrees, we know now why it fails at 200 degrees, and this is very important for the development of the future adhesives because if you know what that your product works, you also want to know why it works. You don't want to stop there because at some point it's not gonna work because we are gonna maybe increase the structure load, or we're gonna say, oh, isocyanates are now awful, like let's not have that many isocycles. You know, you never know what a future holds.

Wojciech Wegrzynski:

Well, that I think that's an interesting question because you if you don't have nitrogen in your fuel, you're not gonna produce cyanides in your smoke. If you have nitrogen there, then you probably will produce that. So so a question of a large-scale impact of that and overall smoke toxicity, for example, out of polyurethane uh glues in timber, perhaps that's a question that will be asked in the future. Uh at this point, I'm not sure if it would be a strong contributor given the little amount and how late into the fire it would go. But yeah, I can imagine such questions being uh brought up someday. Perhaps it's a valid research question. Um, jumping back from um lamination to to char fall off. In your paper, you defined it that it can happen below the bond line, at the bond line, above the bond line. Uh, have you seen any outcomes of using the different types of glue in the progression of that? Any trend that with HPX again it happens way above the the bond line?

Antonela Čolić:

So that's really relevant. So that was our latest finding that we now submitted a paper, it's under review. So what happens is that once you have heat-induced delamination in both adhesives, it is very different what happens later. So in HBS, you have this lamination, and then the char lamella starts to peel off from the surface, and you get really, really soon the char fall-off at the bond line. Whereas for the HBX, the delamination happens, but the crack propagation and the crack itself is not as wide. The bond line still sticks together, and the charred fall-off happens outside the bond line. It happens before the bond line. So you still have some charred layer that is later protecting your bond line. So the reduction of the cross-section is much slower for HBX than uh HBS.

Wojciech Wegrzynski:

Well, you you said it yourself based on the TGA and the TG that the the HBS would uh soften and HBX would char. So I assume HBX would uh like become a different type of char within the char, but it's still like the the the carbon matrix is close enough to each other, so they they kind of form a continuous thing. Well, that that's a fantastic explanation because again, I've seen this and with my own eyes in the experiments we've done with OFR years ago on on the slabs that were under uh thermal load, and and we have seen a difference, quantitative difference between the char fall off between uh slab that was built with HBS and HBX. On the side note, then Quick had a great idea. Let's put a beam inside of the slab, and it changed everything. Uh point where I've started to regret myself being a fire engineer because we had some stuff sorted out, and then we put like one little element inside, and that's where it means everything's changing. Yeah, but but yeah, there's uh there's a World Timber conference paper on the effect of the beam as well. Um I think we've covered most of it. Please tell me how now this translates to your real-world projects. Uh, did your engineering practice change given the findings of your PhD and your colleagues? Did they incorporate any of this in their reasoning when discussing with clients about the CLT timber?

Antonela Čolić:

Well, we are more careful about the specification of the adke itself. There are also some other things that we are recommending, such as lamella thickness or uh having the two lamellas in the CLTs which are spanning in the same direction. And then if we are uh introducing uh the temperatures in our model models, we are internally discussing what the temperatures should be, depending on the on the consequence class of the project or the building we handle. And in essence, PHV also offers you like you cover a lot of statistical methods to go through your data. So later on when you're doing your work, that that is really really helpful when you're analyzing the data that is out there, you can just focus on your own work, like you have to incorporate the world outside of you. So yeah, uh it's you have a wider view on the problem.

Wojciech Wegrzynski:

Yeah, yeah, I think you you've also mentioned you have to care about the where the CLT is used, kitchen office, a swimming pool, I guess. Uh, do do you also take that into consideration? How much moisture will there be in the everyday use of the timber? Is this a factory consider right now?

Antonela Čolić:

Well, to be honest, I've not worked on a project that had such a various uh disparity in the moisture in itself, but I'm sure if such uh project occurs, that that's something that we should consider, definitely.

Wojciech Wegrzynski:

One last question cost action. Was it fun?

Antonela Čolić:

Oh my god, that was strange. Like I was just so lucky because it started when I my PhD started and it finished when it's finished. It's like a four-year thing. So, like they have this thing called STSM, it's like short-term scientific mission. So they essentially funded my travels to Australia and my travels to Canada because Felix Wiesner was one of my supervisors. So I kind of followed the guy across the world. So we worked together in at the UQ uh with David Lang and Sergio Zarate and Thomas Bravo, which are such a great team. And I got to work in that institution and then I did two types of experiments there, like the micro-scale ones, the TGA stuff I did there, intermediate sheer lab stuff I did there. And then I managed to go to Canada to work there with Felix, where I did my survey on the loads that are involved in practice. So essentially it ended up being three research publications, three chapters of my PhD, and also like learning from the other people because they have these like cost meetings yearly or or two times a year in different countries in Europe, where you gather and you present the work and you get to know what is the robustness, how does that affect the design, how does the durability affect the design, what is the influence of the moisture, and you realize that it's not fire only that matters. And it it was just such a great idea, led by a really good group of young researchers. It doesn't matter whether they're young researchers, so it was it was it was really good and very international, which I I love.

Wojciech Wegrzynski:

I got a little exposure to that cost action, it was called called Cost Action Helen. Uh, I'll probably drop a link into the the show notes. Uh I mean i I was surprised with the amount of work out there, so I eventually like uh did not contribute that much, but I remember some meetings uh with I remember you from the beginning, yeah, yeah.

Antonela Čolić:

In the beginning meetings you were there. Come on, do yourself some credit.

Wojciech Wegrzynski:

I I I I was there, but I was like more like slowing down. People are writing their drafts. But uh I remember Ian uh doing great work, Chamit doing great work. Like it was really I I enjoyed it thoroughly, and I'm I'm really happy there are finally tangible outcomes of that work because I know just uh a book chapter has appeared out of that work, and I assume there will be more coming.

Antonela Čolić:

So yeah, it was just published like this week, so yeah, great timing. Yeah, great timing.

Wojciech Wegrzynski:

Yeah, so I I just wanted to bring this up because you know a lot of young researchers are listening, they're wondering what are the possibilities. I'm not sure everyone is aware of cost actions and what kind of possibilities those actions do, but I think they're very, very good ways to get into the international research in the general scope if you're interested. And I also, if if you're listening to this and you're now curious about the the cost actions, I had an episode with uh Nieves Fernandez Anis, and she was also a participant of Cost Action, and actually in that interview, I think we spent uh like 15-20 minutes talking about cost actions in general. So, yeah, there's some there's a lot of uh testament about uh uh about testimony about uh how good cost actions are, and I would recommend that. Antonella, uh well, thank you. Thank you so much uh for uh uh coming to the Fire Science show and highlighting so much about the bond line failure.

Antonela Čolić:

No, you're well welcome. I enjoyed it. This, of course, yeah.

Wojciech Wegrzynski:

I I'm I'm a little worried like Gordian is gonna be so happy about this episode. I just want to say, in no way, in no way Henkel did affect this podcast episode.

Antonela Čolić:

Yeah, Henkel they no Henkel was they were great contributors, especially Gordian. Like that guy is just so helpful and open to discuss. He was genuinely curious about whether the product works.

Wojciech Wegrzynski:

The same the same experience when we were doing that. So this was absolutely like that there was no product placement in this podcast episode, just honest uh scientific discussion between uh two researchers. I felt I need to make this statement after all the praise we gave. But anyway, Abdullah, thank you so much for coming.

Antonela Čolić:

Thank you, thank you for your time, Poi Chick.

Wojciech Wegrzynski:

Uh that's it, thank you for listening. Surprisingly, a lot of things happening at this uh bond line interface. For me, the surprise or new finding was that it's not just you know a single line that that's in between the timber, it actually affects the timber, and uh the blue line and timber form some sort of composite together, which it's not just line, it it is deeper and it kind of makes sense. It makes sense that it works, uh but it also like makes difference to how I uh understand the failure at that place. Yeah, and then the last research was definitely a deep dive into the subject and there's like further reading if you need some more technical details. We always scratch the surface at the fire science show, and there's always more waiting for you if you need that. I think for your engineering practice and you know just the basics, we've covered most of the important things, but if you really want to understand the differences between the HBS and HBX glues, well you have to read more. And finally, it's it's interesting that you know those differences are fairly small, but yet uh enough to impact the large scale. Also, like I I need to learn more about putting more lamellas in the same direction, same orientation, and that that's probably something I would like to explore for the future episodes because I find that interesting. I guess this helps you get your load bearing capacity in the correct direction that you need it more. And uh something we've not tested yet so far in large scale, I think. So I'm probably interested in that uh in that aspect of the design. Anyway, that would be it for the comprehensive adhesive episode of the Fire Science Show. I hope you've learned something new today, and I'm really happy we were able to share this exciting PhD journey of Antonella Cholica. And yeah, I hope we will be able to share more stories like that. I'm super, super happy to host especially fresh PhD students, or fresh doctors actually, fresh fresh PhD graduates in the fire science show because I find those people to be extremely knowledgeable about the area they have just finished uh researching for many years. Therefore, they usually are the number one source of information in very specific areas of the fire science, and I would love to cover more of them. So if you know someone who is willing to share their PhD journey and explain to me what they have found, I am more than keen to listen. Anyway, that's it. Thank you for being here with me this Wednesday and next Wednesday. Uh well, another Wednesday, another fire scientific episode waiting for you. Thank you. Bye.