In the early ’80s, universities around the world were pursuing research projects to expand the knowledge and applications in numerical modelling — The University of Minnesota, University of California, Berkeley and the University of Toronto, to name a few.

Around the same time there were vast improvements to computers. Compared to the slow, complicated main frame systems, the new personal computer created the possibility for applying the advancements in numerical modelling knowledge made within the academic world to the computer programs being written as theses by Master’s and Ph.D students. Over the last 30 years there have been dramatic advances in both computer technology and the software for numerical modelling. Not coincidentally, interest from the tunnelling industry, among others, to take advantage of the potential cost savings has grown in this time. Tunnels speaks with three software companies to discuss the advantages and growing demand for their product. The key, they say, for gaining the most benefit out of the technology is education.

Rocscience
At the University of Toronto in the civil engineering department, professor John Curran specialised in numerical modelling. He attracted a group of graduate students who worked in this area and were very good at looking at the existing numerical methods, building on them, and in several areas, coming up with new methods that took advantage of the new power of the personal computers and the increasing speed and file sizes that could be handled. Their research resulted in several programs, originally written in the old DOS computing language. Evert Hoek was also in the department during a five-year appointment and added his vast experience and expertise to the research at the university in numerical modelling, particularly in tunnelling.

The group, loosely know as "The Rock Group," came up with a number of numerical modelling tools as computer programs that began to be noticed in the industry. There were more and more requests for these programs, which were basic and free at first. They gradually became more sophisticated with features and modelling abilities being continually added, mostly at the request of commercial companies who were now using them both in Canada and internationally.

It was in 1996 that Rocscience was born, a spin-off company from the University of Toronto, so that the software development could be focused on in a commercial environment. "The most important influence of this new company at that time was our skill in creating very good interfaces for these new programs that were the result of research in Curran’s group," explains Evelyn Cunningham, president of Rocscience. "With the improvements and standardisation in business of Microsoft Windows platforms together with the growing PC market, now engineers could begin to use the software themselves."

Before the Rocscience software began to gain a place in engineering offices, numerical software was difficult to use and often required a numerical modeling specialist with a Ph.D. "Rocscience brought this software to the masses," she says. "We didn’t invent Finite Element of Limit Equilibrium for slope stability; we made these methods practical and easy to use."

Itasca
As an offshoot of the University of Minnesota’s civil and mineral engineering department, Itasca Consulting Group, Inc., began in 1981. Head of the department, Charles Fairhurst, had connections through the mining industry who wanted assistance on different types of underground excavation, and decided to start a consulting company.

Roger Hart, principal software advisor for Itasca, had graduated from the U of M at the time, and was the first employee hired. "I think when we started, Dr. Fairhurst wanted to have primarily a consulting focus," he recalls. Later, in 1982-83, Fairhurst contacted Peter Cundall, based in England, who had his own company focused on geotechnical engineering and underground construction, and was developing software to do numerical analysis to support design of underground structures.

"A lot of that work was focused initially on seismic analysis. He developed a program that, in those days, ran on a main frame computer that he used to essentially to do seismic studies," Hart says. Fairhurst convinced Cundall to come Minnesota and later to Itasca, bringing his software including, FLAC, a 2D continuum analysis program and UDEC, a 2D discontinuum analysis program (a distinct element program).

"Peter had developed those two programs on his own and at the time microcomputers were becoming very popular and we worked an arrangement to essentially move those programs on to PCs," he says. Itasca’s customers began to show interest in buying their own copies of the software and a second aspect of the business soon developed. David Russell, Itasca’s chief software architect says, "We feel we have a better way to represent how soil and rock behave up to and beyond failure than other software [such as finite element].

We can give you a less conservative design, but a design that is still stable. Whereas other software is more conservative."

One reason for the increased acceptance of the technology from a civil standpoint stems from the need for design to be less conservative. Hart explains, "Tunnelling construction has to recognise conditions more accurately than, perhaps, they in did the past where they made very conservative assumptions about how material behaves, more empirical type assumptions. In today’s world that is too conservative."

Today there is a huge demand for design optimisation, explains Augusto Lucarelli, a senior engineer with Itasca.

His background is in civil engineering and he has worked as an independent consultant for 16 years on behalf of contractors and governmental agencies.

"Every dollar or every Euro actually makes a difference in winning or losing a project.

"The big advantage that I have been able to exploit from numerical modelling is that it really allows you to have a better understanding of the system without pretending to be ‘right’ (because you have to be aware of what you’re doing). But it brings you a big advantage over more conservative or classical solutions where you really overdesign."

For example, saving 50mm in a final concrete lining or 2m in overall length really can make a difference among competitors. It’s probably one of the biggest reasons driving demand for companies to acquire the technology internally, he emphasises.

Model Myths
A common belief among inexperienced users simplifies the process incorrectly — that is, plugging in information and hitting a button will give an answer: the exact stress that will be encountered.

Russel says, it’s just not like that. "You run a model to determine processes, systems and mechanisms. You don’t run a model because you think that’s actually going to be the exact stress you see or the exact displacement you’re going to see in the result. You’re trying to explore the system."

Lucarelli adds, "A model is a simplification of the reality. You don’t have to pretend to describe the physical world, which is impossible. So every assumption and every simplification brings along some uncertainty and you have to be aware of that. But you have to be able to catch, let’s say, the most important feature of the physical system that you’re trying to model. That’s the real know-how that you have to have in numerical modelling.

"It’s not a matter of running a code without understanding and crunching numbers."

And that takes a lot experience, Russel adds. "If you don’t have the experience you need to be careful."

Rocscience’s Cunningham agrees, "There needs to be an ongoing process to educate the engineer on the dos and don’ts of numerical modeling. The software tools are now so easy to use, and easy to misuse," she says.

Itasca encourages users to work with the company when they are setting up, and can show a recommended procedure to solve specific problems. It also suggests attending training courses and is developing online training options.

The future
The first version of PLAXIS 2D became available in 1987. In the following years, several large tunnelling projects in the Netherlands accelerated the development of Plaxis’ tunnelling capabilities – the first "tunnel designer" in Plaxis was released in PLAXIS 2D version 5 in 1993.

Jasper Van der Bruggen, Plaxis’ sales manager for North America, based in Houston, Texas, explains the most notable tunnel construction project at the time was Amsterdam’s North-South metro line under city’s historical centre. This consisted of a bored tunnel and four stations at considerable depth (more than 30m), with high water levels and poor to moderate soil conditions. Construction was further complicated by the building pits being very close (<5m) to historical buildings with timber pile foundations.

The North-South line accelerated the development of the first Plaxis software going beyond 2D, and this ‘3D Tunnel’ program was released in 1998 and has since been widely used in tunnel engineering.

He says, "With computers becoming stronger, Plaxis calculations could get more complex and faster at the same time. Stronger computers allow for smaller time-stepping, more elements, more calculation nodes, more constructions stages, and therefore for even more accurate results."

"State of the art numerical modelling software offers the tunnelling community the possibility to model complex tunnel geometries using full and true 3D modeling (i.e., not simply extruded in third dimension).

"The sequential boring and/or excavation can be realistically modelled with accurate soil and rock behaviour using nonlinear advanced constitutive models. Specialised software solutions make it possible to accurately and efficiently model the intereaction between tunnel lining and surrounding material, and to correctly model tunnelling features like jet-grouting and rock bolts," Jasper says.