Bentley Focuses on Tightly Integrating Products

Bentley Systems has recently launched a 3D City Model campaign that builds upon a number of their platform technologies, and their long history of working in three dimensions. V1 Editor Matt Ball spoke with Bob Mankowski, senior director of software development, about the company’s geospatial strategy and tool set. Mankowski was recently promoted to a senior director position that oversees product development for the company’s civil (road, site, rail), geospatial (Map, Descartes, etc.), water (Haestad product line) and geotechnical (gINT) products.

V1: Your new role sounds like an interesting challenge with such a broad overview of so many different tools and application areas.

Mankowski: It’s really exciting. We have a lot of moving parts and a lot of people. From a personal perspective it’s very rewarding, because I get to learn a lot. The goal is to bring product development more closely together, and we’re talking to our users to learn exactly what they need to streamline workflows and solve their problems.

We’re integrating Map technology into our civil products like InRoads and GEOPAK. The digital terrain model technology is another good example of our integration. The geo and the civil teams have been working together to take what was civil technology for surface modeling, but making it scale to city-size data sets. And we’ll begin integrating the gINT geotechnical products into both civil and geo products.

The Haestad products are more aligned with the civil products now, integrating the hydraulics and the hydrology in those products and into roadway design tools.

V1: Bentley has been a company that has grown through acquisition, and that seems particularly true within your group. Did many of these products have existing relationships with Bentley before they were purchased?

Mankowski: gINT and Haestad didn’t have an alignment with Bentley at all prior to acquisition. Haestad didn’t have a Bentley development strategy, in fact we were and still are Autodesk and ESRI development partners. The common thread in terms of acquisitions isn’t that the companies were in MicroStation development; the common thread is infrastructure.

gINT does a lot of beautiful reporting and graphing of geotechnical data. The reason that it’s interesting to Bentley is that there’s ground upon which the infrastructure will sit, so geotechnical information is of ubiquitous need across all these projects.

gINT is a database management system for sub-surface data. Soil type is an example data type, where they drill a borehole and geologists determine the types of soils and rocks, and at what elevations these different strata occur. But it goes way beyond that to the structural property of the soil, the water quality and the soil quality. The soil may be contaminated with oil plumes or volatile organic compounds. You can take all that lab result information and tests, and manage that within the gINT product. Its real strength is the ability to query and present that information in reports and graphs.

Talking to our users we’re working to integrate this technology into other products in order to make our whole offering greater than the sum of the parts. Now as part of the Bentley team, we’re working to accomplish things with gINT that they wouldn’t be able to accomplish on their own. They have access to a lot more technology to add value to their products.

V1: We are really interested in the things that are taking place on the visualization front. The ability to display different levels of details was demonstrated on the main stage, and that isn’t an easy problem to solve.

Mankowski: What we were able to do there was to leverage our knowledge and experience with raster. We’ve dealt with huge datasets of raster information for some time, and to be able to display that in a high performance way always involved “tricks.” I’m sure you’re familiar with the idea of pyramids and multiple resolutions of the images so that when you zoom down you see more levels of detail. We were able to apply that same technique to digital terrain models.

There are some unique things about digital terrain models, but we were able to use our experience to cross over and apply it to DTM. That is another example of our civil and geo integration. The DTM is primarily civil, and the raster is primarily geo, but with the two we were able to do something unique and innovative.

V1: Point clouds are an interesting data format that cross over both civil and geo communities.

Mankowski: LIDAR is one technology that is really changing the way we measure our world. We’ve spoken to some surveyors who haven’t done much with LIDAR, because they have struggled with the huge data output. That’s one of the challenges that we’re addressing at Bentley, working with the point cloud and bringing it into the platform to measure off it, draw off it, and snap to it, and do it in a high-performance way. The amount of data is phenomenal, with a small point cloud being something like 3.5 million points, and going up to billions of points.

Using mobile and aerial LIDAR to build 3D city models is interesting, but it doesn’t necessarily replace photogrammetric techniques, in the sense that there is still a lot of value there.

V1: The 3D city models are getting very interesting, where they get to real applications that make a difference, such as IBM’s smarter planet initiative. I think it’s a huge opportunity for the geospatial community as a whole to be at the hub, and to spur evolution of the technology space as a whole. From an infrastructure perspective the model is both above and below ground, right?

Mankowski: The below-ground assets can’t be ignored, and are a really big part of 3D city models, because with a city a large amount of their infrastructure is buried. In fact, not unlike buildings, after you’ve constructed it the best record of it is the documents you used to construct it. Once you close up the ground, or put the pipes and mechanicals behind walls, it’s very hard to know where things are.

The things that we’re doing here in our Applied Research booth with augmented reality are just incredible. You have your as-built information in a form that you can bring into the real world to overlay your digital reality with the physical reality, and it’s going to be huge.

V1: On the city side, do you see a lot of people embracing your full vision or are they still struggling with what it means. There are such divisions between CAD and GIS that seem to still need to shake out.

Mankowski: That does still need to shake out. We do see people embracing the vision. It’s still early, although people have been doing this to some extend for a long time. In other words, some cities have had digital cities for some time, but at the fringes. I think it’s becoming more of a mainstream need.

One thing that I think will drive it is the need to comply with regulations, and the need to have a 3D model to do that. For most purposes today, 2D information and design can get you to the point of compliance with regulation. When 3D information and analysis is required to comply with regulations, we’ll see a lot more adoption, but still, we’re seeing a lot of adoption happening now in Australia and Europe regardless.

V1: It’s getting to really serious 3D now it seems, where the model helps manage infrastructure rather than just a model for real estate or tourism.

Mankowski: Absolutely, and you hit on a key point. I mentioned in my presentation this morning that we’ve all seen 3D buildings in Google Earth, so what’s new about our offering? The level of detail that’s necessary for engineering, analysis and design, and all these types of workflows require this level of detail and information that is different from the use case of digital tourism or advertising and other applications such as Google Earth and Bing Maps.

That level of detail is very different from what Bentley has in mind, and what cities need in terms of managing their infrastructure. Yes, they’ll need models for consumer uses, but you’re not going to do engineering, analysis and design based on those models. You need a different type of model.

V1: I was excited on the main stage with the representation of the 3D model where a 2D drawing is just a slice of the model, without a real distinction between 3D and 2D as separate systems and outputs.

Mankowski: Why force that division to be there? It’s artificial to force it. Why not have your 2D call-outs in your 3D views? That technology is called dynamic views, and what we’re doing with it is primarily model documentation. You navigate the model, and create dynamic call-outs, and it’s not a tedious process to create the merger of 2D and 3D views.

V1: Do you see that technology being adopted in the geospatial realm?

Mankowski: Absolutely. When you look at level of detail, it can start as blocks of building shapes down to the full resolution building model. Cities will have to look at the different levels of detail that they want to support. The dynamic view technology plays right into that.

I think it’s also very interesting for our civil technology as well. There is an artificial division being created in the civil space by others saying that 2D is totally out. A lot of state departments of transportation still require 2D plan sets, because that’s how jobs get done. You may supply 3D models for things like machine control and quantity take-off, but 2D is still a big part of these workflows. The dynamic view and model documentation are really important for civil in this context.

V1: Is there an emphasis on time saving or streamlining of the process?

Mankowski: It’s time saving, but it’s also an enhanced way of communicating with the data. Traditionally, when you look at a 3D model it’s missing all the call-outs, the descriptions and annotations that provide context and the extra level of information to understand what you’re looking at for both design and project review. We can document the model in-place and in 3D, and link it off to the 2D set.

Another example of the blend of 2D and 3D is in our redlining workflow, where you’re redlining a 3D model, and in the past you kind of needed to be an expert in 3D to get into the right view plane to place your text and details. For markup and review, we can now take a 2D snapshot of the 3D model, and manage that markup workflow through ProjectWise.

V1: The different interfaces and workflow-oriented developments at Bentley are interesting. Do the tools themselves facilitate a collaborative workflow or is that something that an organization largely defines?

Mankowski: It’s both. A manager might define a specific workflow, but there are a number of different workflows based on the job. Something that I’m familiar with is in the electrical network area where there are specific workflows for residential service verses commercial service. Where one might only have a single level of review for residential vs. multiple levels of review for commercial projects.

The software definitely facilitates workflows, but they are unique to organizations and some configuration has to take place to set it up in the ProjectWise workflow system.

V1: It’s great to see that the workflow integrates with schedules, there’s a good degree of analysis in place, and there are multiple checks on the design through the process. While this level of attention certainly isn’t new in the work, it seems to be new in the software.

Mankowski: It’s certainly been getting enhanced over time, but it’s not an entirely new technology. The software helps the people that do the really hard job of creating infrastructure by making it easier for them.

Over time, more and more data is captured, and these information models are retained. That builds upon itself. As we gain more information about these assets, and as we build more assets with this electronic information from the outset, future maintenance and enhancements become much easier.

V1: I like the idea of the campus as a group of like-purposed buildings that could even include the plant work that you do. It seems to define a group of activities that incorporate all of your products, and pose the greatest integration possibilities.

Mankowski: Absolutely, the campus encompasses our civil products, the utility, hydraulics and hydrology, roads and buildings, and everything is there. A campus, in a sense, is a mini city, and an airport is even a city within a city in many ways. They have all the same problems as cities in terms of managing spaces, roads, utilities, etc.

I hesitate a little when we talk so much about 3D Cities, because it places the emphasis on municipalities. 3D City encompasses it all, and certainly those technologies are applicable to campuses of any size.

V1: I’ve been impressed with Bentley’s work to integrate more analysis, and was really interested in the genetic algorithms that are now part of many of your products.

Mankowski: The genetic algorithms are the same technology that we’ve been using in our Darwin line of products. We use the technology for three things within our Haestad products. The first is calibration of the model, to get it to simulate the results that you see in the real world, adjusting the inputs so that the results of the model match the results you’ve captured in the field. This tool calculates a huge number of options that are possible, based on a combination of inputs, and runs through scenarios to find the best fit. We also use this for design of water systems, finding the right pipe sizes or the right actions to take to enhance the hydraulic capabilities and at the right budget, uncovering tradeoffs based on potential results versus the expense. We also use it for scheduling, to find the optimal pump schedule.

That same technology is now being applied to buildings and structures. It’s not just an enumeration to calculate the different combinations. Each calculation is very intense, with each solution that you want to try taking a good amount of time to run the calculations. With the genetic algorithms, it’s a way to search through the solution space to find an optimal design. You may not find the global optimal, but you will converge upon a near-optimum solution.

There are other optimization techniques, but genetic algorithms have a lot of benefits. Other approaches sometimes get stuck on sub-optimal solutions, but genetic algorithms have a way of jumping past the sub-optimal to move more quickly to the optimal solution.

V1: I’ve been reading about system science lately, and with complex systems we often intuit the absolutely wrong thing to do. It’s not common sense, because there are too many inputs. Are we moving more toward machine learning in order to tackle these complex issues?

Mankowski: Absolutely, and I think a good example is for water quality analysis in water distribution systems. It’s very difficult to simply look at it, and figure out what will happen to water quality. There are so many variables, that even when we look at the water quality results it might not look right. It’s very counterintuitive, and it takes some time to understand.

For example, it might be because a certain pump turned on at a certain time, and got fresh water in a certain part of the pipe, and the old water out of the tank. You’ll have a slug of old water, followed by new water, and it might mix with other water at pipe junctions. It’s so complex that you absolutely need a hydraulic model and water quality model with analysis software to solve these problems.

V1: One last area is sensors and the sensor web. How important will it be to get more toward real-time inputs for models?

Mankowski: We’ve seen a number of things going on in that area. In our applied research team, we’re looking at the real-time sensor web. It’s absolutely something that we recognize as having importance for geospatial. There are plenty of reasons to be able to incorporate real-time input into models.

We’re seeing interesting uses for 3D city models with real-time feeds from video cameras. Being able to take the movie and project it into the 3D model allows you to see the city more in context. These capabilities aren’t within our software today, but we’re seeing them as a very interesting part of future development of the 3D city.

We’re focused a lot on the engineering and analysis parts of the equation, and all sorts of sensors can inform a lot of our analysis workflows.