Addressing the Productivity Crisis in Construction

JAIMIE JOHNSTON: Hi, all. My name is Jamie Johnston. I'm a board director at Bryden Wood who are an integrated design consultancy. And I'm going to be talking about addressing the productivity crisis in construction. So I'm sure most people have probably seen this slide. It was published as part of one of the McKinsey reports a few years ago. And it shows how productivity has dramatically increased in almost every other sector apart from construction.

And quite rightly, construction gives itself a hard time about this. And it's definitely one of the things that, as an industry, we need to address. But there's something quite interesting in this graph. If you look at how they actually measure it, it's the value add per employee. So you take some raw material, you process it, you turn it into a product, and the cost difference defines the productivity.

So certain sectors have had a massive leg up because traditionally, things like oil and gas or pharmaceuticals couldn't help but be highly productive. They took relatively low cost feedstock, processed it, turned into things that are much more expensive. And so they've traditionally had a massive leg up in terms of productivity. So we don't quite have that benefit.

So whilst there is definitely enormous benefit in being more productive and McKinsey in one of their other reports said there's a potentially 5 to 10 times boost available if we could become more like manufacturing, there's certain things that mean we can't simply copy manufacturing. And construction will probably never look like manufacturing. So I'm going to explore some of the blockers and what are some of the things that we've been doing, certainly in the UK, and increasingly, North America now, to try and shift into this much more manufacturing style production system as McKinsey call it.

So the first or obvious question, why aren't we more like manufacturing? So industry has been talking about this for decades. There are definitely some blockers that we need to address at first. So the key one is repeatability.

So I'm sure everyone knows, every construction project typically is a prototype. So we form a new design team. They develop a design from first principles. We deliver a project. If there's any learning, it doesn't typically get disseminated because that team gets disbanded and move on to another project.

The first thing that automotive did to really get the hang of this was get the hang of repeatability. So prior to Henry Ford and the Model-T Ford, every car was a handmade, luxury, bespoke item, much like construction. The key thing that Ford did was say, well, if we made them all the same, suddenly productivity gets much higher because repeatability of processes, of components, of the whole system is what's going to start to unleash this.

The first thing automotive did was get the hang of repeatability. That's going to be absolutely crucial for us. This is a mental picture that lots of people have of the future of construction. So I think people tend to imagine we'll replicate manufacturing. We'll have six-axis robots making components. I'm not sure that's ever going to be quite true.

So there's an interesting thing, I think, in manufacturing that typically the process is bigger than the products. If you're making a phone or a car or something, you can have a process which is bigger than the component you're making. Buildings aren't like. That they're typically much bigger scale.

We're not going to have 200-story high, six-axis robots making buildings. So there's definitely a place for automation, but it's probably not as much in the factory space as certain people think. The obvious exception to that big in the process or big in the product is aerospace where a Boeing 777 is the size of a building, but it costs about $200 million.

So the other blocker is cost density. So buildings are mostly air and commodity materials. If you had a cubic meter of iPhones, it would be worth about $14 million. A cubic meter of concrete is about $150. There's about 100,000 times more cost density in an iPhone than concrete.

And that's quite an extreme example. But again, it tells you something about what we should be doing. So trying to move everything into a factory, if you take a very low-cost material, handle it a bunch of times in the factory, and bring it to site, you've actually massively increase the cost without necessarily increasing the value to the end user. So these sort of combinations of trying to be more repeatable, working out what to place in factories and what not to put in factories, and how to get automation properly on a construction site to manage these things-- that's where we think this is going to head. And that's why probably construction never quite looks like manufacturing.

So the approach that's been taken, particularly in the UK and particularly driven by central government is to stop thinking about individual programs or projects and start thinking much more broadly. So I talked about that repeatability point earlier-- the fact that every sector focuses on why it's special, why it's different, that's a huge blocker. What we started to do here was say, well, if you look for commonality across sectors, started to identify common kits of parts, that's how you might start to address some of these things and how you might start to make this shift.

Probably the best example of an existing platform is IKEA. So everyone knows when you go to IKEA, the interfaces are the same. The process is always the same. The book's always the same. You only ever need a hex key and a screwdriver.

So IKEA has taken all furniture, it's boiled it down into a very small number of repeatable components and processes, and it means that whatever you're making, whether it's a bed or a chest of drawers or a bookshelf or a kitchen cupboards, you're always using the same components, the same process, the same tools. And thus, that's transformed the furniture industry from something that was quite bespoke, quite handmade, quite trade-intensive like construction, suddenly, they completely deskilled it and almost anyone can make furniture now.

So the platform thinking was, what does that look like for construction? Could we get down to small number of components and processes and start to massively repeat them? So this is the terminology the UK government used. I did a talk on this at AU last year. I'll put a load of links in the class handout that have all the previous documents that we've spoken about.

But the key principle was this idea of looking for commonality. So to give you an example, lots of buildings have about an 8-meter span. That's because, at the height of the person, you can get natural light about 8 meters into a building. And that's why schools and healthcare wards and apartment buildings have quite common structural spans. It's nothing to do with sector. It's everything to do with people.

So the thinking started to say, well, if you could identify more of those repeatable features, it would reveal a kit of parts that would reveal the things that you could start to do across sectors. And suddenly, you get into the sorts of numbers that manufacturing likes. That's how you start to make this big shift into a much more manufacturing-like industry. If you had that, you'd then have a library of digital components that could be configured by design teams on the right-hand side.

If you had the rule for those components, you could probably automate that. So on the right-hand side, you start to get into autoconfiguration. You start to get into higher productivity in the design phase without necessarily relinquishing any kind of site specificity or quality of design. On the left-hand side of this diagram, you then have your physical components, which could be manufactured by a broad diverse supply chain, assembled by teams of people.

So it's not mechanics or engineers who make cars anymore. It's people who have been trained to assemble components. Again, you start to deploy automation, and this is how you start to get a more productive process through the design, manufacture, and assembly process, again, without reducing the quality of the design while still getting quite site-specific assets. So we've been developing a number of these platforms now-- the numbers we use just to describe the kind of spanning characteristics.

So platform naught, which is our low-cost, rapid accommodation. Platform 1 is cellular accommodation for things like cells for the Ministry of Justice here. It's for single-living accommodation for the army, student accommodation. Platform 2 is our mid-span, up to 8-meter platform that we've used for education, healthcare, residential. Platform 3 I'm going to show you an example of is the next scale up. It does a 9-by-9-meter grid.

Platform 4 is for much bigger span-- sheds for distribution warehouses, those sorts of things. But one of the key things for us is that we use the same components and the same processes across a whole range of these. So I talked earlier about IKEA using the same components for beds and chests of drawers and wardrobes. We use the same components for schools and healthcare facilities in offices.

What it means is as soon as we've learnt something in one sector, we can almost instantly deploy it in other sectors, which means all of these sectors are moving much, much quicker. All of the learning from one project gets built back into that digital kit of parts, physical kit of parts, and improved upon for the next project. So this is how we get into that continuous improvement cycle that manufacturing's been so good at.

So one of the things that's allowed manufacturing to keep getting better and better is the accumulation of loads of incremental gains. So every component gets slightly better. Every assembly process gets slightly better. And over decades, that adds up to enormous improvements in productivity. So I mentioned that repeatability thing earlier. This is why we've never been able to get into that continuous improvement because construction has typically constant reinvention.

So by having a more stable kit of parts, a more repeatable set of assembly processes, this is how we start to really over time improve these. And this is how we start to move learning between sectors very quickly. And this is potentially how we start to shift the entire industry.

So I'm going to show you what this looks like in practice. Again, I've spoke a little bit about this project last year. So this is an update. The project is nearly complete now.

It's an office block, commercial office, that we've done for a client called Landsec. They're the biggest private developer in the UK. This building was actually designed traditionally, initially, got planning consent, and then we retrofitted a platform kit of parts to this building. So you can see here, it's very, very site-specific.

So the quality of the design is very responsive to its context. The site is actually between Tate Modern and The Shard if you know London. It's opposite St Paul's Cathedral on the south of the river. It's a very high-profile site, lots of very high-profile neighbors and stakeholders. So we couldn't have a design which was not high quality on this site.

So I think that's a sort of proof point for platforms that we've, again, not compromised the quality of design, but we've used a kit of parts that we've developed in other sectors, applied it to this building, and I'll show you some of the metrics we're getting out of it. So we broadly focused on superstructure, facade, mechanical electrical systems, and fit out.

You'll see for us, we've got superstructure in the middle of this diagram not because it's the biggest cost element, by any means, but it's the enabler of these other things. So you can already buy perfect millimeter perfect facade systems. You can already buy prefabricated mechanical electrical systems.

What we were seeing is that the benefit of that prefabrication gets massively diluted if you bring a millimeter perfect thing to site and you have to interface it with a traditionally built superstructure, which can be plus or minus two, three inches in terms of location of columns and things. So what we were seeing was lots of these potentially very prefabricated systems were never achieving their full potential benefit because of the time taken to measure and fill in the gaps between the manufactured thing and the traditionally built thing.

So we said, if we can get the superstructure to be super quick, super accurate, then it's really believable that everything else kind of installs the kit of parts. So again, it's exactly what automotive does. So they take a single sheet of metal, it gets pressed in a huge stamp. It turns it into the car chassis, and that's then the structure and the setting out and the carrier frame for all these other components that then get assembled onto it.

So if you think about the superstructure as your car chassis, it's a carrier frame for all these other systems. That's where we put the most effort, and it enables all these other things that are already happening out in the industry. So this is obviously time lapse of the building going together. What you'll notice here is firstly, the site's incredibly constrained.

Central London site-- there's virtually no laydown space. And so having this prekitted kit of parts. Think of your IKEA wardrobe. When it turns up, it's in a box. You can take it straight into the house. You don't build it in the garden or the garage. You actually take your kit of parts to point of view. So logistically very controlled site. You can see the speed. Obviously, sped up, but you get a sense of how fast these components are coming together. And again, this is our first iteration of construction as manufacturing.

So to take you through those key components, the superstructure, the facade, the MEP, I'll show you how we assemble them and I'll show you some really interesting metrics that started to come out from this project. So the superstructure was actually 90% of it, 80% of it was using a kit of parts we originally developed for the Ministry of Justice.

So I talked earlier about this idea of transferring things between sectors. We developed a very lean kit of parts for Justice primarily for their mid-span buildings of their education buildings, their healthcare buildings, some of their offices, added one component to get these bigger spans. And we've already taken a thing which developed for the public sector, given it to the private sector who have now picked it up and refined the design and have moved it forward.

And the next thing we'll be doing is handing back all of the benefits from the Landsec project back into the public sector for the next iteration of public-sector buildings. So we took a kit of parts we'd previously developed. I mentioned earlier that digital library of the components. So because we had all of the components already modeled, in many instances, we didn't actually need to draw them.

So we were able to send our files directly to the supply chain. So this is a place where I think robotics has a part to play. It's not in the construction of entire assets, but it's probably in the fabrication of these highly repeatable components. So we literally sent our files to the supply chain. They were laser cut, laser welded, robotically welded, which means that they're submillimeter accurate. They're incredibly precise. And again, this is how we start to get this accuracy that's currently lacking in construction.

So then on site, this was our prototyping facility, but this is how we deliver the building. We were then looking to see whether we could take the construction process and, again, like IKEA, turn it into a very simple set of very repeatable tasks. So you'll see here very few people. We're measuring things in minutes. All of these assembly tasks are very, very straightforward.

They're all very repeatable because they're using laser cut components. They're are all super accurate. And this was looking to see whether we could install the entire superstructure from the floor below. So if we could build a superstructure without ever needing to work at height. So slip, trips, and falls are still one of the biggest health and safety risks. We thought if we could remove that and turn the whole process into something you do from the floor below, you get much safer working, much more productive working, and you can see here with a handful of people, a little bit of automation, not six-axis robots, these are pieces of equipment that are common in the distribution warehouse logistics industry.

We've just appropriated that technology. And it shows that you can potentially build a building half as quick with half the number of people. We also prototyped it. So you can see this was our prototyping center. So again, a thing that manufacturing does very well, it tests things, iterates things, refines them before it puts them into a line environment.

So by the time we arrived at the construction site, this process was well understood, the metrics were well understood. We knew how to put these components together very, very quickly. Some of the benefits that came out of that because of the repeatability of the individual components, we could put an awful lot of effort into optimizing each component and the system as a whole.

So one part of that was we took out an awful lot of materials. So there's 30% less embodied carbon in the building. Because of the prekitting, the logistics was very, very highly controlled, as you saw from that time lapse. And the overall impact was, for this project, we took out 30% of the carbon. It's the first commercial building in the UK to be certified by the UK Green Building Council as net zero in embodied and operational carbon.

And we're on track to get a five-star rating on the Neighbors UK scheme. So enormous sustainability benefits. And again, you imagine multiplying the scale of that benefit by the scale of the UK public sector. You can see how this approach could dramatically address some of our big climate crisis issues.

We then had a team led by Dr. Danny Murgia at the University of Cambridge. So we had a number of PhD students crawling all over this project. So they had access to the site eye. They had access to all the Gantt charts, the day records of what happened, people's time sheets. So they were able to break down what happened and start to really dig into what happened on site and what could have happened on site.

So normally, of course, on a construction project, there's so much noise it's very hard to actually see what's happening and definitely to tell what could have happened. The enormous benefit here was because we have this repeatable kit of parts and we have these repeatable processes, we can actually measure things quite accurately. And we could see in quite a lot of detail what was happening on different floors with the different activities and actually how the project was unfolding as it got built.

So of the two buildings, you can see here that's the visualization. Block A was the bigger building. We focused on that because it had the biggest chance of getting people into a flow of activity. So these are the average rates we got on this project. So the productivity, about 330 pounds an hour. Install was about half a square meter per operative hour in terms of the superstructure.

So we had a benchmark of typically how the project was performing, which was probably slightly better than traditional construction. Not surprising, particularly, given that this is the first project where some of these things have been brought to bear. So we're expecting a learning curve. We're expecting to at least do as well as traditional construction, but hopefully, show that if done consistently, this approach would have dramatic impact, which is what we found.

So looking at the individual levels of the building. Level three was one of the bigger levels. It was the second level using the proper platform kit of parts. So the initial learning curve has been achieved. We're starting to get consistency.

The floor plates actually get smaller as we go up the building, which is why the numbers drop. But level three gave you an insight into what was very, very definitely achievable with a bit of practice. The really interesting thing, it showed that had that level of productivity being maintained, you could potentially have built the building 40% quicker. So you could have got a potentially 40% increase in productivity not by going quicker, not by doing anything much better, just by consistently getting those rates.

So we know they're achievable because they did them on one of the floors. What it shows is that you need to get consistency. And suddenly, there's potentially massive impact just ready to get your hands on. So again, this is why we think the platform approach has got a potential long life because you can see here that just on the first project, we're getting these numbers. If we kept doing this and got better and rehearsed it more, these numbers would get much, much better.

And it also showed you that one of the key things that got in the way was variability. So you can see here in terms of putting up the steel, the best day they had, they put 12 pieces of steel up. But on seven days, they didn't put any up. So again, it shows you don't have to be quick, necessarily. If you were just consistent and kept on your average rate, actually the boat would go faster, the whole building process would go faster.

Found the same thing in the common floor, which is the down stand beams. Again, massive variability between 13 days where none went up and one day where 10 went up. If they'd just maintained at a average rate of five a day, you could have been more than twice as quick in terms of delivering these things. We get the same thing in the shuttering, which is the stuff we pour the concrete onto.

Again, best day, 18 pieces a day. Number of days where none happened. If you just maintained the average, you could have taken out 26% of the program. The other way that Cambridge visualized this data for us was in a thing called flow lines where they show installation of components and the gaps in between is where nothing happened.

So these are the days where either they were off doing other activities or they ran out of components or the logistics team hadn't kept up. But you can see for the primary steel, about a third of the time was inactive and for the common floor, about a quarter of the time. So again, normally, it's impossible to see these sorts of things on site. Suddenly, we've got a real insight into the measures that we can take to improve productivity aren't necessarily about going faster. It's just now about can we plan those activities better.

Now we know how long they take, can we ensure that the teams are geared up to do them? Can we plan the building around these install rates? And potentially, you get much, much better results.

And the other thing Cambridge did was they looked at the way the phasing on the building worked. So they said, actually, if you had rezoned the building into smaller zones with teams that cycled quicker, without doing anything quicker, you could have built the building potentially 43% quicker. So this range of numbers from 25% to 40% is definitely readily achievable if we'd known this ahead of time. And obviously, the benefit of platforms is now we do know this. You can plan the next projects around this.

MEP, we then, having installed the superstructure, it was designed that all the bays were the same. All of the fixing points for the MEP were already cast into the slab. And then we worked with the supply chain. And this time a company called NG Bailey. And said, if you knew that the superstructure was designed in such a way that all the bays were the same, that there was this massive rationality, what could you do about it or how would that improve your processes in the factory?

So they actually set up benchtops with the same adapter frame, they called it, on every bench. They could drop the MEP mechanical electrical systems onto those frames, stuck up a series of cassettes. These aren't volumetric cassettes. This is actually three, four, five, six cassettes with a little spacer.

And they set their entire factory process up around churning these things out. They put little wheels on them, and then they literally wheeled them out of their factory onto the truck, wheeled them onto the floor plate, and then on the right-hand side, you can see we had specially adapted forklift trucks that would pick up five or six or seven of these things at a time, lift them up. The fixing points are already there. So they staple them in, move on to the next one.

And literally, a single operative or a couple of operatives, again, a little bit of automation, not complex automation, not robotics, but an adaption of an existing piece of equipment, and suddenly, you're getting much, much higher levels of installation rates. So suddenly, they're moving much, much quicker. And it looks incredible. It looks like a machine inside.

So we're able to integrate the MEP and the superstructure very, very closely integrated whilst designing in all the maintainability, and the flexibility, and the ability to adapt things in the future. Because we have the fixing point grid, it means that we've already got latent fixing points for the next generation of MEP whenever they need to refurbishment. And because it was so closely integrated with the superstructure, it meant that the floor-to-floor heights in this building were quite low for very good floor-to-ceiling heights.

So that reduced the entire building volume, and therefore, the amount of air that we need to heat and treat, and therefore, the running cost, the operational costs, the operational carbon. So all of these things start to contribute to a leaner, more sustainable, more material-sensitive building. And you can see in the foreground here, that's a series of those cassettes ready to be installed.

When you walk around this building now, it feels incredibly well-organized. It feels very integrated. They're exposing the surfaces. And you can see here the impact of having that equipment.

So to start with, on the left-hand side of that graph, that's industry best practice using genie lifts to move these cassettes around. Once we designed and adapted the forklifts, you can see what happens to the install times. They plummeted from 150 minutes in some instances by at least a third some. Things came down by 90%. So you're suddenly into much, much more productive working-- one or two operatives doing large areas of MEP.

And normally, a lot of the package plant would be prefabricated. The risers would be prefabricated, maybe the main horizontal runs. But typically, with mechanical electrical systems, once you get away from those, you start to do things trade by trade. Because of that rationality, we're able to do these multi-trade cassettes, do most of that work in the factory.

And you can see here, Bailey reckoned that took out 30,000 hours of site labor. And that's work at height, and it's work in quite painful conditions, working overhead. So that's a dramatic improvement in the working conditions of the people on site who absolutely love this way of working.

The final key element for us was the facade. So as with the rest of the building, we designed the facade as a super rational kit of parts. It wasn't a special facade in some sense. We went out to the normal facade supply chain.

But because of the rationality of the design and because they could see that the superstructure would enable them to work better, they actually gave us a cost reduction. So they said as a facade manufacturer, 40% of your time on site is spent in measuring and shimming and bracketing and masticking and-- we talked about earlier-- filling the gaps between your millimeter perfect unitized system and the traditional piece of concrete that sits behind it. So they immediately gave us a cost reduction simply because they could see they wouldn't have to work around it.

On 9-by-9 meter bays, we were at plus or minus five millimeters on the superstructure. So that's a level of accuracy which is very common obviously in things like manufacturing. It's very uncommon in construction. But it enabled the facade supply chain to give us a fantastic deal simply because they knew the facade would fit. It would be much easier to install.

They actually got down to seven and a half minutes of panels. The original program showed seven panels in a shift. So that's about an hour a panel. And at peak, just before Christmas, they managed to do 90 in an hour and a half. So that's like the difference between someone changing a tire using a pump jack and taking 45 minutes and a Formula One team turning up and zipping these things on. So that when they did the 19 in an hour and a half, they then had to leave site because they ran out of panels.

So it tells you that now these rates are achievable, logistics has to be able to keep up or you'll keep diluting the benefit. And that was one of the key things for us was now we know these rates are achievable, firstly, could we improve them. Secondly, let's make sure the logistics and just-in-time and the delivery is there to support that kind of level of working.

And again, same thing. We got the same sorts of results. And the variability was quite large. If they'd just kept their average, you could potentially clad the building nearly twice as fast. But again, no one ever believed that at the time. And so logistics just weren't in place to support that speed of working.

So final couple of slides for me. So what it shows, what the Cambridge data shows is that without even going quicker than we currently did, if you just got that consistency, if you could just deliver day by day by day the same sorts of rates of work, this number keeps coming up. There's a 40% reduction, 40% improvement in productivity readily available.

So you don't have to do anything better at this stage, you just have to get that consistency. And I think that was mirrored by talking to people on site. People on site wants to have productive days. They want to come to work and they want to know what they're going to do. They don't want to have these days where they're hampered by lack of materials or lack of information or lack of tooling or whatever it is or they can't get into their work zone.

So there's something interesting here around there's a way of building buildings much quicker that's also much more satisfying for the people doing it and gives them a much better way of working. And it's really easily achievable. We don't need to do anything much better at this stage. If we just got consistency right, then potentially, we could be building things much, much quicker.

So key things for us that we're looking at next is logistics. You can see here actually Gantt charts aren't as much use to us as we thought. Now that we've seen this Cambridge data and the piece count and the flow lines, you could start to plan the entire building like that. You could start to get into discrete event simulation and using all of these tools that manufacturing does to get this continuous improvement.

So first thing for us is we could now start to plan building in terms of component numbers and assembly times and really start to plan things much more carefully and therefore support the logistics. And consistency is much more powerful than speed at the moment. So we don't need to get quicker, but if we could just be consistent, that would be the first big leap.

Future steps that are coming, as I said, the learning from this project has been fed back into the public sector. The learning from this will go into things like our new hospitals program here in the UK. We're already getting interest from clients in North America, in US and Canada, as to how to start to adopt a platform approach over there. So there's nothing in this approach which is in any way UK-specific.

The idea was always, if these systems work, they work in any geography. So the potential is there for the industry to really pick this up, to run with it, to propagate it more widely and really make that transformational shift into something like manufacturing. And what we're hoping we're going to see over the course of the next couple of years is as we start to get more into this, as more programs and projects start to use it, we'll work out more about what to use in terms of automation.

Clearly, once you've got into that space are very repeatable tasks, you don't need trades in quite the same way. Not that we're going to automate people out of jobs, but some of the big challenges facing construction are skills gap and aging demographic. So as I'm sure most people know, in the next 10, 15 years, a big chunk of our skill base is about to retire. And we're not getting enough young people coming in.

The idea that you could turn construction into those very simple, very repeatable processes means that we could potentially train people very quickly. And we've actually built pharmaceuticals facilities using ex-British army. We've built bits of the prisons program using prisoners. We built bits of airports in the UK using people from a local job center.

So this is one of the reasons governments are really interested in this idea that we could set up training and competencies. And suddenly, our construction labor force is comprised of ex-unemployed, ex-services, ex-prisoners. We resolve the skills gap. These are the sorts of things that I think are going to make a big, big impact in the industry in the next few years.

There's definitely a place for logistics in all of this. So current construction logistics gets things pretty much designed pretty much at the right day. We're miles away from that just in time manufacturing-type mindset. But you can see things like the facade. When they do 19 panels in an hour and a half, you want to get the next set of panels turning up in the afternoon.

So suddenly, logistics becomes super important. And it's going to be the biggest next barrier that we have to overcome. And yeah, I've already talked about this, but the idea that all of the learning from this gets put back into the next iteration of components, puts back into the next iteration of the digital library. We've really got a repository and a place for all this learning to go.

So yeah, we're very excited about this. Hopefully, that's been interesting and hopefully we'll be back next year to talk about the next steps and how the platform approach is continuing to move apace with some examples from North America. So that's all for me. Thanks for your time. Thanks for your attention.

As I say, there will be links in the class handout to the videos, to the animations, to documents we published on this. Hopefully, it's a topic that will keep expanding over time. And hopefully that's been an interesting half an hour. Thank you.