Digitizing Tunnelling and Geotechnical Engineering with Rajat Gangrade

Andrea Ciaffi

Hi Rajat, welcome to the Digital Construction podcast. How are you today?

Rajat Gangrade

Very good. Thanks Andrea for having me over.

Andrea Ciaffi

So today on the Digital Construction podcast, we dive deep underground, really literally and digitally because we are here with Rajat Gangrade. He is a technical advisor at HNTB. He's also Co founder of ground up. It's a startup. I'm very excited to learn more.

Andrea Ciaffi

About it and Rajat also previously worked at Arup at Acom, so he has a lot of experience and at the moment he's in North America, in San Francisco and very excited to talk about all of the aspects of tunnelling and geotechnics. But from the digital point of view, so this is a very exciting.

Andrea Ciaffi

Episode today. So to start today Rajat, I would like to talk about this concept of digital first approach for tunnelling and many of the people that don't work in

Rajat Gangrade

Mm hmm.

Andrea Ciaffi

the infrastructure sectors may not be familiar with it. But can you share with us a little bit of what are the aspects that can be digitised in the context? Of tunnelling and infrastructure projects.

Rajat Gangrade

Yeah. Interesting question. I typically like to break the projects into different phases in different cycles that the project actually goes through. So during the design and planning phase, I think a lot of emphasis is on digitising the archive data.

Rajat Gangrade

Be it your technical data or be it you know the data from desktop studies, and those are typically archived or digitised in the sense in in terms of building preliminary digital twin models or or preliminary. Visualisation models that exist for the planning phase and BIM is a good example of it.

Rajat Gangrade

You know you as an engineer want to visualise the utilities and the structures and what environment your project is typically set up in. So this this planning phase really helps with understanding class detection with the utilities and and I don't know what sort of obstructions we might be facing in in terms of alignment optimization.

Rajat Gangrade

That's the first phase. The second phase is more towards the detailed design phase where you know engineers take a deep dive into what exists and they start building tools that in compass, bringing in GIS aspects and then bringing in.

Rajat Gangrade

The actual footprint of the structures along the tunnel alignment and bringing in the 3D geological model, so it's it's basically creating a single source of truth or a common data environment for different data sources. Those are some of the striking or I would say key.

Rajat Gangrade

Features when it comes to digital aspects and tunnelling, but the most interesting part comes during the construction where we have actual construction data being read by sensors, be it the tunnel boarding machine or be it the data from instrumentation and.

Rajat Gangrade

Monitoring all of that integrated into, you know, maybe a digital twin or single piece of software and all this, all this data is transported to an environment through cloud based solution. So that's another key striking feature of.

Rajat Gangrade

Of of digitalization during construction and the last part of it is the operation and maintenance part, where once the structure is complete, we want to create an archive of the project in a way where it becomes easier. To track and record any predictive maintenance and incident logging.

Rajat Gangrade

Sometimes we do a very sophisticated version of inspection of the tunnels using automated tools or or rewards nowadays. That capture the interior of the tunnel as singular images, and then there are machine learning algorithms that they utilise to understand, you know which part of tunnels are experiencing some sort of water leakage or crack detection.

Rajat Gangrade

So that helps with, you know, the automated documentation of the as built conditions of the tunnel and then all of that is, is is typically stored with this project stakeholders. So yeah, I mean digitalization has its own.

Rajat Gangrade

Use case in every phase of the project depends on what part of the project we are we are in, but I think it's it's. It has a tremendous amount of application during the construction phase of tunnelling.

Andrea Ciaffi

It's great to hear all of the opportunities we have and one of the things I've learned over the years working this digital space for the construction industry is that different regions may have different level of maturity.

Rajat Gangrade

Mm hmm.

Andrea Ciaffi

And I'm based in Australia. I'm in Sydney. You are in San Francisco and North America. So I'm just keen to know from your point of view, how digitally measure the tunnel in sector is in North America.

Rajat Gangrade

Yeah, that's that's a very insightful question. While. From what I see, I think Australia is way ahead of North America just to just to mention that I think in terms of when it comes to digitalization during construction, there are a few contractors over here and consulting companies over here in the US.

Rajat Gangrade

That are, you know, pushing the envelope on adapting digital technologies to advance their practise. However, I would say I would still rate the maturity level to be relatively low and it's lower than I expected. To be honest with you, in the era of 2025 and you know over the over the recent years there is good amount of awareness.

Rajat Gangrade

However, the practise in North America, I think faces some aspects of barriers as well. The conservative industry culture, there's quite a bit of inertia on adapting the technologies. There is an uncertainty on, you know, what is the cost benefit relationship once these technologies are adapted and applied so.

Rajat Gangrade

I would say it's it's, it's an open canvas at this point in time in North America and it presents an exciting opportunity for engineers like us to propose, build and basically prove what these solutions can add to projects.

Andrea Ciaffi

Yeah, it's an interesting space because it there is a lot of barrier in this industry to adopt adopting digital

Rajat Gangrade

All right.

Andrea Ciaffi

technologies, but yeah, it's it's exciting to see how this evolves over time. And one of the. Things I'm always curious about, but you are an engineers 1st and at some point he must have decided. Let me focus on the digital stuff and I'm just keen to know how did you develop this interested in digital rather than just applying the traditional approaches that were available for your discipline.

Rajat Gangrade

Yeah, it was actually during my PhD when I started exploring these tools, you know, it was for the first time in in any aspects of my career, be it professional or even personal, that I started to advance. I started to code and I started to write scripts and to automate most of my work that I did during the PhD and.

Rajat Gangrade

It was a massive learning curve in the sense I mean, as as a civil engineer, you're trained to ask the questions, how and why, and there are codes and standards to refer to, and then you know, you go to those and you find references and and then blah, you just make it work.

Rajat Gangrade

It's such a different field when it comes to digitalization and you know when you adapt and start coding and scripting, because there are tonnes of resources you need to philtre out the right resource and adapt the right optimal approach to get to what you want to do.

Rajat Gangrade

But thankfully there are tonnes of resources available online as well, so that really helped me. I saw examples of how automation is performed, how digitalization aspects of tunnelling are done. I mean there was quite a bit of limited amount of research, so most of it was my own imagination of if I want to visualise some aspect of data.

Rajat Gangrade

How should I do it if I want to see the statistical distribution of a property? How should I do it? And I took a lot of inspiration from other fields and other industries. I would say medical science was one of them because there's a lot of statistical analysis that happens in medical science as well.

Rajat Gangrade

So I looked at those publications. I looked at those plots and then I used to figure out, oh, can I replicate similar things for geotech and tunnelling industry? I also since since I started coding I also traverse this. Fine line of, you know, developing an application of web application or a software and then also being the engineering brainchild behind it.

Rajat Gangrade

But it it really gets very challenging. You can either be a software developer for the application or you can be. And engineering product manager, I would say so was confused for quite a bit amount of time, wasted a lot of time doing that, but I think it was an iterative process and my PhD really helped me gain that level of.

Rajat Gangrade

Expertise and confidence to to develop and to implement such digital tools.

Andrea Ciaffi

Yeah, sometimes it it comes from the desire of solving a problem in a slightly different way, and digital is probably like the the easiest thing you can find in front of you. And and I love your approach of looking at other industry or other other aspects where of course we can learn a lot from other industry that has solved.

Andrea Ciaffi

Of similar problems they have. Maybe they have a mindset that they have

Rajat Gangrade

Yeah.

Andrea Ciaffi

used to solve the problem. That's also great. It's something I've done a lot as well, so that's great.

Rajat Gangrade

Yeah.

Andrea Ciaffi

I would like to talk about data because one of the things that I think are very

Rajat Gangrade

Yeah.

Andrea Ciaffi

important at the moment that it's data and. Data is the foundation for all of the things we do. We talk about data ownership, but from from your point of view, think about geotechnical and tunnelling. What are the types of geotechnical data and tunnelling data that are most often?

Andrea Ciaffi

Overlooked or underused?

Rajat Gangrade

Yeah, I think when it comes to, I think one of the most. Most often under looked and overused, and this may sound counterintuitive. Is the data from geotechnical side investigations. I was going through some of my own work and and some of the latest research on.

Rajat Gangrade

The failure incidents on tunnel projects and sort of came across this statistic last evening that talked about that about 77% of projects in underground construction and tunnelling this this suffer. Because of insufficient data or an inadequate interpretation of the data, so I think.

Rajat Gangrade

And and what happens is we we do both we we we perform investigations but that that is such a low number in terms of percentage wise of the volume of the soil excavated that it leads to a significant amount of uncertainty. And on that limited number of boreholes we carry out further limited testing, analysis or laboratory testing.

Rajat Gangrade

So that basically reduces a sample space. And you know there is no spatial analysis of the data in that sense. There are no, you know, no, no details about you know what? Where should we drill more boreholes just to understand and reduce this uncertainty,

Andrea Ciaffi

Hmm.

Rajat Gangrade

I think that becomes a huge challenge. So I feel that the investigation data that we get from from site investigations is is definitely overlooked and and needs to be investigated a lot more. There have been projects, as I mentioned in in. In North America and worldwide, that have suffered because of inadequate ground conditions and not utilising the data completely.

Rajat Gangrade

I think I I think that remains to be one of the biggest challenge in terms of utilising the data.

Andrea Ciaffi

And from your point of view, how can we improve the way we manage data to to then overall to improve the outcomes of the the projects that we work on. For example, if you're working on tunnel projects, how can we improve our way to manage the data to improve the actual outcome of the projects?

Andrea Ciaffi

Yeah.

Rajat Gangrade

Yeah. I mean, I think it it, it all comes down to geotechnical workflows and that is what our industry has struggled with for a significant amount of time and for a long time it is. It is probably going to be a very fundamental conversation when I bring in the topic of workflows to maintain, organise, analyse and then interpret geotechnical data currently.

Rajat Gangrade

Most of the industry practise is is involving using fragmented or siloed data sets and we struggled with version controls of the files. We struggled with analysing the data in bits and pieces. It imposes a significant challenge to to the engineering community, a smarter geotechnical database management system is required.

Rajat Gangrade

Basically, that is that serves as an umbrella to every single. Step of the workflow that that we need to adapt, be it math and logical operations, be it visualisations, be it statistical modelling industry, must they must look into a system of of of that.

Rajat Gangrade

You know that starts serving. As a common data environment or single source of truth for engineers like us, when we see a comprehensive picture of the data into a single environment, it helps us understand what the gaps in the investigation are.

Rajat Gangrade

You know, and then obviously once you get access to smarter, leaner data, you would utilise the data to prepare leaner proposals and then win projects and build, you know obviously utilise this as legacy applications into advancing your business.

Rajat Gangrade

So overall, I think along with Geotech database management, an aspect that is picking up some steam over here in North America is is the integrated digital delivery and you know it's not just maintaining the data in a database, it's not a database problem.

Rajat Gangrade

It's more of a problem of integrating different sources of data into one single environment and you know getting away from A-1000 page PDF to a single screen where you can where you can visualise and you can see what the data is all about.

Rajat Gangrade

So I think for me that is that is what geotech database management should be like.

Andrea Ciaffi

Yeah, it's a good point. It's all disturbance are struggling with the trying to integrate the different data sources from different platforms into one single place. That's a difficult, difficult thing to solve at the moment.

Rajat Gangrade

Right.

Andrea Ciaffi

One you touched previously on the uncertainty rather to. The specific order of your discipline of geotechniques, so you sometimes you have to work with unknown and you have to think about the

Rajat Gangrade

Yeah.

Andrea Ciaffi

variability of the space and the again the uncertainty. So how do you because you may have unpredictable surfaces you may have done. Samples on in specific places, but you don't have all the information

Rajat Gangrade

Mm hmm.

Andrea Ciaffi

which doesn't like from engineers really don't like uncertainty too much, and digital tools also don't like uncertainty too much.

Rajat Gangrade

Right. All right.

Andrea Ciaffi

So how do you integrate this uncertainty into your design process?

Rajat Gangrade

Yeah, that's a great question. That was my PhD was all about. So I think the idea that we are not able to excavate 100% of the ground and Despite that fact. We are fitting in an engineered system into the ground for a period of 120 years.

Rajat Gangrade

It just amazes me it just the kind of risk that we take as engineers to, to excavate tunnels and and and build the systems is. Is phenomenal. So when it comes to integrating uncertainty, I think it's imperative in the 1st place and 2nd there are robust approaches as I would say probably.

Rajat Gangrade

40 years ago someone what someone would have done was change one number and see what the change in output is. And you know that is that is a very fundamental way of doing things. You change something. You see what the output is, you change something again and then you change the combination of parameters and then you see what the output is.

Rajat Gangrade

Later on this was advanced a little bit to something what we call as Monte Carlo simulation nowadays, which is using a statistical distribution and you do 1000 iterations of it and you expect that the true value of a parameter would just lie within that.

Rajat Gangrade

Distribution. And at some point in time, you will hit the true value. You will get the output, but you would also get a standard deviation on the uncertainty and that is how uncertainty is quantified or uncertainty is integrated into design procedures. My thesis talked about more sophisticated approaches on using.

Rajat Gangrade

Stoke stochastic algorithms that you know that quantify the variability mathematically and capture the pattern of variability of geotechnical data. Variability measures such as standard deviations, inter quartile ranges, those are. Extracted and those are then translated into into the determination of parameters. So it was more around looking at what the statistical distribution of the data was and you know how.

Rajat Gangrade

Mathematically, the data varied from one point in space to another point in space, and this sort of created the opportunity to build probabilistic approaches or probabilistic methodologies. Where we have multiple possibilities or realisations of the ground conditions and what would that lead to in terms of capturing the uncertainty in a parameter?

Rajat Gangrade

In in a, in a very simplistic way. As an example, if there is an uncertainty in a geotechnical parameter that affects the tool, where on the tunnel boarding machine, then we would be able to translate that uncertainty into the risk of tool. Where and basically see what is the uncertainty in the tools wearing out eventually.

Rajat Gangrade

So it's all about capturing the variability in the soils which is, which is just inherent because of the way the geological depositional processes take place.

Andrea Ciaffi

And do you use any specific tools to make informed decisions with uncertain data?

Rajat Gangrade

Well, yeah. I mean in for there are there are a bunch of tools that available. Geotechnical industry has over the years relied upon using autocorrelation functions in my phdi started to use geostatistics. That was based on stochastic algorithms to to derive and quantify this uncertainty and then translate it to quantify the level of risk.

Rajat Gangrade

Another level of tools that I would think that would be used to communicate uncertainty to project stakeholders. And helping them make informed decisions would be. Engaging, compelling visualisations. Just imagine having a visualisation with a heat map of a certain risk projected along the plan view of the tunnel alignment in an urban setting.

Rajat Gangrade

That just gives a clear indication to the stakeholder on you know what areas of the project are experiencing risk or what may not. So I think visualisations, they convey the probabilistic measures and help a user intuitively understand. The results with the threshold of risk obviously being defined these visualisations.

Rajat Gangrade

As I said, they can be confidence intervals, they can be heat maps, they can be cumulative distribution functions. They can be probability curves, but yeah, I mean uncertainty and spatial variability. I think both are conveyed better through visualisations than just quantitatively.

Andrea Ciaffi

Yeah, definitely. And yeah, you can see how the different aspects of digital can help solve problems, but also inform stakeholders, which is great because sometimes even digital to even to manage expectations of

Rajat Gangrade

Yeah.

Andrea Ciaffi

the different parties involved, it's it's key.

Rajat Gangrade

Right.

Andrea Ciaffi

One aspects we when bads were there, I should say it's automation we we've used this automation word for a long time now in the industry and that's real like things can be automated, processes can be improved.

Rajat Gangrade

Right.

Andrea Ciaffi

I just wanted to ask you.

Rajat Gangrade

Yeah.

Andrea Ciaffi

From your point of view, if you've seen any examples of automation into techniques or tunnelling.

Rajat Gangrade

Yeah, the word automation really gets me pumped up because it is it is. It is such a low hanging fruit in in terms of, you know implementing and and and tunnelling is it just presents a significant amount of opportunities. Although being a low hanging fruit, not many have thought about advancing the industry practise in the sense of automating routine tasks.

Rajat Gangrade

So I think you know examples of this particularly includes bringing about. Efficiency in design processes or includes speaking to software using codes and scripts or allowing 1 software to speak to other software. And then there is interoperability of data or even. You know something as basic as building robust repositories and design libraries that contain a variety of calculations.

Rajat Gangrade

Something that I have come across at Arab before, where most of the design calculations are stored in libraries and they help automate the routine calculations on every single project. But another example that I came across. Fairly recently, I think from the Australia Tunnelling Society was you was using numerical modelling tools like Plexus and flak using Python And is this is just amazing how?

Rajat Gangrade

One level of scripting is helping run a lot of cases to characterise and quantify the uncertainty in the tools in in, in, in the in in the results, so I think. That example of automation was really phenomenal. I would like.

Rajat Gangrade

I would like to personally implement it at some point in time and as I as as as you must have already guessed it, you know my interest lies in building or developing these library of solutions for analytical calculations. I believe every. Company, every contractor, every community, every everyone part of the community should be advancing the industry in that direction.

Rajat Gangrade

You know where the human intervention is minimal. There lies a quite a bit of challenge in maintaining, you know, the appropriate data structures and then. QA QC for people, senior professionals who are not well accustomed with automation. They would struggle with the QA QC aspect.

Rajat Gangrade

So it needs to be bared in mind. But again, I think in the current state of practise, you know automation should be directed more towards bringing in data from different sources and then automating your your typical regular workflows. And when I say different sources as in.

Rajat Gangrade

An example of automation could be, you know, geotechnical parameters, project constraints, tunnel geometry, what is the impact of tunnelling on nearby structures, so that really helps paint a very comprehensive picture in terms of automating our workflows and automating our tasks.

Andrea Ciaffi

Play song. Yeah, I love your approach of trying to work together, all of us, to improve the industry and with automation routines. It's something that this podcast is also trying to do, trying to share knowledge. So we overall we will achieve a better

Rajat Gangrade

Yeah.

Andrea Ciaffi

outcome for for everybody rather than developing and improving.

Rajat Gangrade

Right.

Andrea Ciaffi

Different in different silos, everybody in isolation. One of the aspects of automation to like I've been speaking about these automation tools and processes and workflows with many people over the last many years and one of the challenges is adoption because sometimes people struggle to see the

Rajat Gangrade

Yeah, yeah.

Andrea Ciaffi

benefits and. And also even the limitation, it's also important to understand the limitation because I always find that there are people that are they don't see the benefits, they don't want to accept that they want to adopt anything. There are also people that are over

Rajat Gangrade

Right.

Andrea Ciaffi

excited and they don't see the limitation. So they expect something incredible. But then they get they get a little they end up a little bit disappointed because

Rajat Gangrade

Yeah.

Andrea Ciaffi

they didn't understand the limitations of the tools. So from your point of view,

Rajat Gangrade

Right.

Andrea Ciaffi

what are the benefits and the limitations of of adopting these automation tools?

Rajat Gangrade

Yeah, I think. Since I have since I started my education for civil engineering, I've just heard two compelling variables that need to be kept in mind. Is budget and time and you know everything needs to be within budget and

Andrea Ciaffi

Yeah.

Rajat Gangrade

time. And I think one of the major benefits of adapting automation is respecting those two variables to a greater extent. It is the biggest benefit is developing these trustworthy deliverables within a QuickTime and you know, it gives us an additional level of flexibility to deep dive for any further analysis.

Rajat Gangrade

Once you have the basic. Deliverable ready. So design automation is is is an exceptional opportunity for a young to mid level engineer to to understand the engineering concepts and and deliver the project efficiently.

Rajat Gangrade

Aspects of design processes like sensitivity analysis, understanding the concepts for a young to mid level engineer it becomes way lot easier with automation because one is just not struggling with making or drafting a new spreadsheet for every single project and.

Rajat Gangrade

And although the whole process just becomes more efficient in the sense that we start focusing on building in or bringing in additional details into our into our deliverable. So I think that's one of the biggest benefit that I've come across. And in terms of?

Rajat Gangrade

Limitations, obviously there is this limitations of you know most of the times it becomes again coming back to budget in time it it becomes a bit impractical to to invest that level of. Capital and and time on projects, especially on projects with tight deadlines and you know heavy workloads, it takes me back to this amazing xkcd comic that I'm using.

Rajat Gangrade

You talks about investing a significant number of hours developing. An automation process to be used only once a year. So it all depends on how many times we are going to use this process and it becomes a bit challenging convincing the senior leadership or senior management in in, in many companies that you know automation can bring about an an efficient experience for them.

Rajat Gangrade

So and obviously there is inertia. Obviously there is conservatism between the within within the industry. Many organisations, they, they they still rely on legacy systems or manual processes, making the transition to automated workflows difficult. So I do see that there would be some sort of inspiration or motivation for this companies at the start of their.

Rajat Gangrade

Financial cycle to adapt to these technologies, but sometimes the cost just drives them away, so that's a that's that's a major limitation of automation. I mean it's something that cannot be embedded into the current scheme of things for any companies, organisation or workflows very easily.

Rajat Gangrade

It's it's, it can't be done like that. There needs to be. A proper vision there needs to be a proper procedure in order to make it most useful to yourself.

Andrea Ciaffi

Yeah, definitely. And when you think about the adoption of this digital tool, sometimes I see teams that want to adopt something because they on one side, they think it's a good idea. On the other side, they want to deliver a better outcome for the projects, but sometimes they don't think about.

Rajat Gangrade

Yeah.

Andrea Ciaffi

The prerequisites of how they they need to be ready to to be able to adopt automation to. But from your point of view, what are the prerequisites so teams and projects and companies can actually adopt automation tools or processes.

Rajat Gangrade

Yeah, that's a great question, the. The question actually leads to the aspect of lacking any standards or guidelines as such on on developing such procedures or workflows and every company will have to identify their like their own prerequisites and. Umm. That, that, that suits them the most, be it within their workflows or be it, you know, creating a robust infrastructure it depends.

Rajat Gangrade

But according to me some of the important prerequisites for implementing automation within workflows is is. Digitization of the data, I think that's the most fundamental requirement, appropriate database management. You know there should be a a standardised process that is developed within the firm.

Rajat Gangrade

That helps them QAQC the outputs for project deliverables. I think QAQC is the part where the automation gets hurts the most because because of the lack of communication or a barrier of communication between the teams. Of you know. All right, you have implemented this.

Rajat Gangrade

How do we check this? How do we check it for various cases? So I think QA QC is something that can be standardised. There should be strategies on how.

Rajat Gangrade

Data will be brought in and how data will be output for for project delivery and these are all housekeeping items I would say but they but they create such a massive difference to to bringing in the aspects of automation within the within. The companies and I think one of the most important one is, is having sustainable, skilled workforce.

Rajat Gangrade

We are struggling with that quite a bit over here in North America. But again I think it is a problem that can be solved. The incoming workforce is. Generationally talented, I would say in terms of adapting digital tools are more tech savvy, so if we need skilled workforce that will be capable of implementing modern practises, there needs to be a multidisciplinary coordination.

Rajat Gangrade

For example, the geotech and the tunnelling team needs to talk to the numerical modelling team or or someone who's doing more automation on cloud computation and then therefore they need to have communication with software developers. So I think the prerequisites are more on.

Rajat Gangrade

Having some sort of standardisation within the firms to to help them make most use of the automation.

Andrea Ciaffi

Yeah, definitely. And at the moment we live in the AI era, so this is the time for AI. Like previously,

Rajat Gangrade

Yeah.

Andrea Ciaffi

we were all talking about automation at some point, somebody started talking about blockchain not so much in the construction industry,

Rajat Gangrade

Right.

Andrea Ciaffi

but it looks like it's the time for AI. It's very hard at this point in time to understand what is the role of AI in the construction industry in general and if you try to focus on one specific

Rajat Gangrade

Hmm.

Andrea Ciaffi

discipline, it's probably even harder to understand how AI can improve the specific workflows. But I'm just keen to know your.

Rajat Gangrade

Yeah.

Andrea Ciaffi

You know your take your point of view on what the role of AI in the construction industry at this point in time.

Rajat Gangrade

Right. Yeah, I think it's it's extremely. Challenging to quantify that because there's a lot of noise, there is tremendous amount of noise on AI and you see stuff that possibly someone has developed a prototype and are and are willing to implement, but then it.

Rajat Gangrade

It works on run project but doesn't work on other project and there is this another aspect of introduction of AI into the industry is through the academia and if you look at the publications coming in from. This universities worldwide, not only in the US but even China, Australia, Europe, there is tremendous amount of noise.

Rajat Gangrade

Everyone is just churning out papers on AI maybe every week and. Those are all opaque. I would say opaque and difficult to reproduce into the industry, so it is the responsibility of engineers like us to determine the role of AI in the construction industry.

Rajat Gangrade

There are tangible benefits and outcomes that are possible. You know, using AI, but it's our responsibility to find such use cases. In my personal opinion, the role of AI is to take away the heavy lifting that we do on our routine task, I mean.

Rajat Gangrade

And does not want to struggle to write once a quick reference to something that needs introduction about a topic or has a hard time understanding, I think we should focus on using AI for these fundamental tasks first. Gain confidence on how things work.

Rajat Gangrade

And and then proceed on to implementing AI for complex engineering tasks, yes and yes, of course. I mean there have been, there have been cases where AI has. I mean there have been cases where AI has been used for complex engineering task and has been successful as well.

Rajat Gangrade

But I think for the major part of the Community it is, it is easier for them to navigate into AI and find use cases just by implementing simpler things. And for professionals who love to code, I think Sky is the limit. I think you know, AI can be used.

Rajat Gangrade

To develop all sorts of engineering applications and and you know, can be embedded easily into their daily routine tasks. So I think I think a biggest part of AI in construction is just automating some of the routine task for now.

Andrea Ciaffi

Yeah, makes sense. And but do you think that the industry is mature enough because we have been debating how they're not the indices,

Rajat Gangrade

Yeah.

Andrea Ciaffi

maybe sometime not mature for things like automation? Do we just jump straight into AI or maybe they are the solution? What do you think about the measure? Of the industry and adoption of the AI solutions.

Rajat Gangrade

Yeah. No, my, my, my opinion would come as come out as very startling or surprising for the most part of the community. But I I don't think industry is especially over here in North America. I don't think they are AI ready. And the reason for that is.

Rajat Gangrade

Again, I mean, there is a lack of standardised guidelines on AI and then no one knows what the cost to benefit relationship is. There is no proven performance metrics on AI, so I sense a strong inertia within the industry to adapt.

Rajat Gangrade

To AI solutions and I wrote about this a couple of months ago where you know and I mentioned previously on this as well that the way the academic research is progressing is is just creating more noise rather than contributing to bringing in practical applications of AI.

Rajat Gangrade

They're doing AI research at the speed of light 10 papers a week on an average maybe. But where is all of these being implemented? I don't see them being implemented. I don't remember the last time an engineer picked up. A general publication and said, oh, this is what I'm going to implement on my project doesn't work that way.

Rajat Gangrade

So a small cohort in North America is attempting to implement AI for tunnel design some famous applications. That have been discussed in North America are on optimising, you know TBMTBM operations and there there has been a significant amount of research being done at Colorado School of Mines and UC Berkeley, the on the University of Illinois at Urbana Champaign.

Rajat Gangrade

On how BBM operations can be made more efficient? Oh yeah. I would also like to mention there has been some extensive research on the same topic in the East Coast of the US as well. Where they have focused quite a bit on you know, understanding the outliers in TBM operations using AI and then building model for structural geotechnical analysis.

Rajat Gangrade

But again, all of this comes as a. Comes on background of some success in academic research, but in terms of practical applications, it's been a big challenge over here in North America at the least.

Andrea Ciaffi

Yeah, it's, it's this. We are at this stage where there's this new thing and we think about how to use it. And yeah, it will be. It's an interesting space to watch at the moment. Do you think that geotechnical and tunnelling? Have any differences or similarities with potentially other disciplines when when we think about the user usage of AI?

Andrea Ciaffi

Is there anything we can learn from other disciplines?

Rajat Gangrade

Yes, definitely. I think we can. We can learn quite a bit from what other aspects of the OR or what other industries are implementing into their into their processes. But again, tunnelling in geotech days. The the industry's so unique and and that uniqueness comes just from the uniqueness in the projects.

Rajat Gangrade

If we are taking an example of other industries implementing a service in AI for a particular problem statement. I think they might be seeing that problem statement at on on multiple projects, but that's not typically the case. Within the tunnelling and the geotech industry, so certain aspects of design applicable in one project may really not be applicable to other projects and we we face that challenge.

Rajat Gangrade

But again, this sets up a perfect opportunity or use case for what the tech industry is calling nowadays. As Agentic AI, you know, which is basically just building an intelligent model capable of autonomously learning and adapting to the change in product condition. Project environment change in machinery.

Rajat Gangrade

Change in construction means and methods change in design approaches. So conventional AI or traditional AI is not the answer for geotech and tunnelling industry. I firmly believe that agentic AI is the future for our industry and A and is and is a much better, perfect fit for us.

Rajat Gangrade

The challenge with that is we do not have any background on it as as as, as as a workforce, it's not been introduced in our curricula. It's it's not been.

Rajat Gangrade

There have been we have not been trained for it as such, but I hope this challenge can be navigated with traditional AI, so we need to utilise the traditional AI as a stepping stone towards building solutions that would be focused on implementing the agentic AI that takes into account.

Rajat Gangrade

The change in project environment and conditions significant challenge to that would be the cross collaboration among the industries to share data. I mean if there is no data, there is no AI. So we really need to work on that aspect as as as industry.

Andrea Ciaffi

It's another example of the technology evolving faster than the industry, faster than the industry can adapt to it. But you are working on an exciting

Rajat Gangrade

Exactly, yeah. Right, right.

Andrea Ciaffi

projects in addition to all the things that keep you busy today, which is a startup, it's called ground.

Rajat Gangrade

Right.

Andrea Ciaffi

Up I'm very, very keen to learn more about it. It's a startup in Geotechnical for Geotechnical professional understand and I just want to know what is the problem that ground up is solving and how did the idea come to life?

Rajat Gangrade

Yeah, sure. Yeah. No, we are very excited about ground up. So ground up is our founder, James Holcomb's idea he was working at Arab and was developing or building softwares for geotechnic solutions. And he came up with this idea of. You know, transforming the geotechnical engineering practise with systematic database management and.

Rajat Gangrade

I think I I met James over LinkedIn chatter with him about the possibilities of building an extraordinary piece of solution and and that's how we collaborated. But ground up is aiming to.

Rajat Gangrade

You know, transform the geotechnical engineering practise in in a much bigger way than it is carried out right now we want to integrate database management or I would rather say we are integrating database management, exploratory data analysis, all statistical modelling. Integrating visualisations, helping engineers go from raw data to insightful data or report ready visualisations in a very short amount of time.

Rajat Gangrade

We have sort of realised that given a period of time in the conventional practise, engineers spend about 70% of their time just maintaining databases and the remaining 30% is typically spent on and analysing the data, interpreting the data, looking at the distribution. But most of the time it's just, you know, oh, I created this file.

Rajat Gangrade

Let's create another file and then, you know, try to more speak make one file speak to another file. It all becomes painful. So we are moving away from that. We want to get the user into a single cloud based web application and do all their math logic.

Rajat Gangrade

Operations into a single environment and we want the user to move from raw data to. I would call as polished output in in no time in in. In you know, with less amount of effort, so that leaves. Good amount of again time and budget to look into the details of of of the data or deep dive into it.

Rajat Gangrade

That is the that is the biggest problem that we're trying to solve in addition to it we are also looking into you know integrating. Aspects of ground modelling and and and integrating aspects of GIS mapping into into ground up. So basically they're building a comprehensive workplace to bring in all your complex your technical data.

Rajat Gangrade

Transform them, transform the data into something very actionable, and something that is report ready for project deliverables.

Andrea Ciaffi

This is certain, certainly an exciting space and 1:00 to watch for all the people involved in this industry. So good,

Rajat Gangrade

Yeah.

Andrea Ciaffi

good luck with this project and care very keen to see how the the project evolves

Rajat Gangrade

Thanks.

Andrea Ciaffi

and every time that there is somebody trying to improve. The technology and the space and the tools for all of us, it's always a great initiative and a great and mob will get a great outcome for you and your team and the rest of the industry.

Rajat Gangrade

Thanks. Yeah.

Andrea Ciaffi

And one of the challenges we have in the industry is culture and change and how to.

Rajat Gangrade

Yeah.

Andrea Ciaffi

To convince people to basically adopt these digital tools, we live in an industry. This is an industry where there is a big variety of people with different ages and there is a younger one that, as you said, they are a little bit more.

Andrea Ciaffi

Tech savvy. There are native digital people,

Rajat Gangrade

Yeah.

Andrea Ciaffi

but we also have the most senior people that tend to be the people that make decisions around what to do in the in the business. So from your point of view,

Rajat Gangrade

Yeah.

Andrea Ciaffi

what are the biggest barriers? To adoption of these digital first workflows, particularly in geotechnical and infrastructure projects.

Rajat Gangrade

Yeah, that's a very valid point that you made about the senior management making decisions and them being a bit less aware than the younger generation that's coming in. So that's really creating sort of, I would say, a barrier in itself. In terms of, you know, different expectations and different, you know strategies on understand and and running the business.

Rajat Gangrade

But I think one of the biggest barrier that I have understood so far is the is the complexity of of technology. And you brought in a really good point of technology is different to different people. And for someone in senior leadership who's been in the industry for a significant amount of time for them, technology could just be an automated excel sheet.

Rajat Gangrade

But for an incoming engineer, technology could also mean why don't we use AI to just read drawings and and get the outputs? So there are difficulties integrating the technological aspects into our businesses. And I think that's that's that's that's really a big barrier.

Rajat Gangrade

The organisational silos as you mentioned, I think poor communication and collaboration across the teams and. Across hierarchies it it creates a significant barrier I mentioned about the huge capital investment that technology requires and you know there is this pressure to deliver the projects in time and on budget it it creates sort of an environment.

Rajat Gangrade

Environment where the teams start to think that we are better away practising or implementing, you know these technologies on a different project and we'll do the digital first workflow some other time, let's just get. This project done and let's just proceed with with our deliverables.

Rajat Gangrade

So I think those those still remain a significant barrier for our industry. I think we need a bit more amount of of patience and and a bit more amount of budget obviously to work through it. Yeah.

Andrea Ciaffi

Yeah, it's always the budget and the projects and the timeline that really affect how digital is delivered, but we're close to we're close to the end and I would like to focus on the future

Rajat Gangrade

Yeah. Uh huh.

Andrea Ciaffi

of digital and tunnelling and what trend do you see are shaping the future of. Tunnelling globally from the digital point of view and if there is anything that that you like and what excites you the most about these new trends?

Rajat Gangrade

Yeah. Yeah, that's a great question. I think as as a digital evangelist slash, an engineer, I'm most excited about the virtual design and construction aspects typically known as VDC in our industry or in the AEC industry. This is one of the top trends that will shape the advent of digital use cases in tunnelling and underground construction.

Rajat Gangrade

So for people who do not know, VDC is basically a project management methodology that utilises. Digital technologies, particularly BIM or building information modelling to plan design, manage construction projects more effectively. It involves creating a digital representation of the actual physical project. What I personally call as.

Rajat Gangrade

Is a digital archive. You're like, imagine once a project is done and you are served the project on a table, but in a very digital way on a on a screen of an iPad. Probably that is allowing stakeholders to, you know, visualise, analyse and you know, optimise all aspects of of of construction.

Rajat Gangrade

I think what makes this unique is this approach helps us move away from traditional methods of using digital tools. You know, and helps us gain the comprehensive story create. A real time impact or or a real time project representation. I would rather say it can answers obviously coordination and there is the aspect of improving efficiency.

Rajat Gangrade

All of that is is being said by tonnes of people back in the past. I'm not saying anything new. But I think it'll get to a point, especially in the tunnelling field. It'll get to a point where integrating a variety of data into a single environment helps us achieve the VDC and it's going to be a phenomenal use case for our industry.

Rajat Gangrade

And that is definitely. Into the future, we want project stakeholders from different disciplines, different backgrounds to come together to work together and and solve problems that are typically solved in silos. At this point in time. So I sincerely hope.

Rajat Gangrade

That engineers like me and like yourself and like everyone else, they they harness the strengths of virtual design and construction in preparing excellent solutions for for, you know, most complex and. Challenging problems. This could rather be like a technical solution to refine the builds or build economically sustainable projects.

Rajat Gangrade

Again, it's an open canvas in engineers like us, we define the. We define the problem statement in that sense, but I think VDC is the future.

Andrea Ciaffi

You know that's that's definitely we really hope for that. And where do you see the industry in the next 5 to 10 years because you're describing a future that we really hope we will achieve. But if we try,

Rajat Gangrade

Right.

Andrea Ciaffi

if we try to stay a little bit more practical and think about what's a little bit more tangible the next like it's it's kind of like a. Short, medium term, maybe for the industry with yeah,

Rajat Gangrade

Right, that's.

Andrea Ciaffi

5-10 years. Where do you see the improvements?

Rajat Gangrade

Yeah. No, that's a great point, I think. I think over the next 5-10 years we are probably going to see a growth worth, I would say 20 years because of all the integration possible, integration of agentic AI and explainable AI and digital twins that would happen.

Rajat Gangrade

There is great amount of awareness and great momentum at this point in time within the within the industry. To understand these terminologies of digital twins and you know, integrating. Geotechnical and instrumentation data real time for real time monitoring. There is good amount of awareness in terms of you know how these tools will help make data-driven decision making.

Rajat Gangrade

So I think over the next 5 to 10 years, digital twins and virtual models are going to really pick up and as an industry I feel we are and I hope that we are going to make significant advancements in adapt, adapting these technologies for sure.

Rajat Gangrade

Cloud based collaboration is something that is going to happen or is expected to happen, at least within the next 5 years. Companies will slowly go through the process of raising the bar on their infrastructure for cloud based collaboration.

Rajat Gangrade

And this would be one of the first steps to take to integrate data from different sources and different softwares and different with different data structures, so that that really is going to be a problem solver for us. And I think the last part is within the next 5-10 years.

Rajat Gangrade

I mean, the digital literacy of of our industry is going to go up significantly. The incoming workforce is hugely talented. They probably are exposed to more opportunities and situations than we could ever imagine. So I'm pretty confident that the incoming workforce would bring in that fresh breath of air and, you know, help us help us empower the digital aspects that are possible within the tunnelling industry.

Rajat Gangrade

So yeah, I mean, I'm pretty excited. For what? What, what the next decade holds? Because I see a significant amount of growth and what has happened in the past decade or so.

Andrea Ciaffi

Yeah. Thanks rajat. This was a great episode.