Michael Barkasi's Blog

Digging Deeper Into DTM

Mike Barkasi — Fri, 01 Aug 2008 06:01:00 GMT

Digging Deeper Into DTM

In our last post we looked at the process of working with PowerCivil. Hopefully you have come away with the reasoning that somehow you will need to get your existing surface into your project. This will provide you instantaneous feedback as the new surface objects tie into the existing ground. We do this by creating a TIN (triangulated Irregular Network) representing the existing conditions.

DTM or TIN?

When working with 3 dimensional digital surfaces you will come across two common Acronyms. DTM- Digital Terrain Model and TIN- Triangulated Irregular Network.

There are many forms of DTM available in the engineering and GIS world. The two most common forms for DTM include Raster and Vector. A common misconception is that a DTM and a TIN are one in the same.

A TIN is a vector data structure which depicts geographic surfaces as contiguous non-overlapping triangles. The vertices of each triangle match the elevation of the terrain exactly. This means that a topographic surface is represented by several triangles, with each triangle face having an approximate slope, aspect, and surface area. The irregularity of the triangles comes from the scattered nature of the (x,y,z) points (the triangle vertices) used as a background elevation source -- hence the name irregular. Once a TIN is created, the elevation of any point on the triangle's continuous surface can be interpolated.

DTM (Digital Terrain Model) is simply a generic classification for all types of terrain models. A TIN is one means of creating a DTM other file formats such as DEM (Digital Elevation Model) could also comprise DTM. Triangles need not be present to have a DTM as the figure below depicts. Still other DTM use Raster grids These DTM do not rely on triangulation but on the weighted value of the grid itself for computations.

Create A TIN

PowerCivil has the ability to create a TIN from a variety of information or a combination of the same information. Below is a quick and dirty list of some of these data formats.

Survey Data
Vector Data (points, contours, breaklines,
ASCII data
DEM
LIDAR (Sonar)

No matter which type of data we use to create the surface the process can be simplified into 2 steps.

1. Extract- This process extracts the X,Y,Z values of the data and stores it in a .DAT file.
2. Build- This process (Build Triangles) reads the .DAT file data and triangulates.

A Perfect World

In a perfect world after those two steps we could move on in our project. Perhaps "Load" the TIN and display it. Or maybe a quick analysis. Unfortunately nothing ever seems to be that easy. Forces outside our control challenge us at every step. Each type of data has its own challenges. Fortunately for us PowerCivil gives us the tools to deal with these challenges and leave us with an accurate DTM to represent our surface.

Extract Graphics

As stated earlier the extraction of graphics extracts the XYZ value of the vertices for use in our TIN. First let us look at some typical types of graphics we work with.

Contours
Triangles (3D faces)
Points
Elevation Text
Breaklines

Contours- 3 dimensional contours pose little problem for PowerCivil. There are 2 main issues we can run into with these types of contours.

1. Proxy Cell Object- Some programs write the contours a what we call a proxy cell object. This basically means the entire set of contours appears as a Cell or block. In this form the extract graphics tools will not work. This does not mean we cannot work with these objects, it simply means we will need to use a couple more steps to render the data useful to us.
a. We first need to "UNLOCK" the data. To do this we can use our "key IN "dialog (Utilities>Key- In). Type in "Change Unlock" select the contour AEC (Proxy Cell Object) and hit Enter on your keyboard. This will unlock the graphics.
b. Next we need to "Drop" (Explode for you AutoCAD people) this will leave us with graphics we can work with.

While this will allow us to work with those AEC objects the exploded contours can become a series of very short line segments (at times ,2' in length) While PowerCivil will work with these graphics they are not desirable. Graphics of this sort can easily create TINs in excess of 1 million points on small sites depending on the number of contours. For this type of site we will pull out some tools made specifically for these processes.

c. Extract Graphics- After we drop the graphics (Step B above) we can extract the graphics creating our .DAT file.
d. Join Linear features- (Civil > DTM> Edit> Join Linear features) this will allow us to re-join all the individual line segments creating a new .DAT file. The cool thing is we can see immediately the number of segments before and after the join.
e. Filter DTM features-(Civil>DTM>Edit>Filter DTM Features) Following the join we will have much fewer line segments that will help in processing the TIN, however we will still have a nearly the same number of vertices on the new .DAT file. Typically I will filter the joined .DAT file by the XY-tolerance of .1 or .2, this will give between a 60 and 90 percent reduction without sacrificing the integrity of the contours.
f. Finally we can triangulate our joined and filtered .DAT to create a workable surface.

The second problem can come in the form of the "Proxy Cell Object (above) or a standard 3D contour drawing. But it involves overlapping contours. We will look at this and other types of TIN hurdles in the next post.

Keep in mind that while we can create the surface from contours these are not the best data to build a surface. In my opinion the graphic triangles (3D faces) work best when dealing with extracting graphics as these are what the original contours are interpolated off of to start.

Getting Started With PowerCivil

Mike Barkasi — Tue, 17 Jun 2008 06:51:00 GMT

Getting Started

It seems like no matter what we do in life; "Getting Started" is the hardest part. Whether we are at our job, home working on a project, dieting, exercise or writing a blog, getting started is the hardest part.

Perhaps it is the fear of the unknown or simply not being comfortable with the task, if you're like me, you avoid things you haven't tried.

There are a lot of people out there totally new to PowerCivil. My first post "Intelligent Design" was to make a case for why we need a tool like PowerCivil. My intent with this post is to outline a process and drawn similarities to your previous work flow.

Before we begin, we will make some assumptions:

1. You have PowerCivil and have it installed and licensed.
2. You know how to create a new file using a seed file. Or you have an existing file DGN or DWG and want to use it to begin.
3. You can create a line or maybe more using Microstation

The Process

My primary goal as a designer was to take an existing site layout; draft a design representing the new use of the site; layout or design storm sewers, sanitary sewers, and potable water; and produce plans and reports.

The process in PowerCivil is basically the same as above. However, if we don't know the tools and are not familiar with the terminology and the process, we are set up for failure.

In a nut shell the process looks like this:

•1. Begin with Existing Conditions

•2. Create a Model (3 dimensional surface representing the design)

•3. Create drainage, and water, and sewer projects

•4. Create plan sets

Amazingly the CIVIL menu reflects this process

Step One - Existing Conditions

Creating an Existing Conditions site plan, or more importantly for our process, a TIN (Triangulated Irregular Network) representing the existing site can be divided into two categories:

1. Create from the actual surveyed field collected data (Civil > Survey >DTM)
2. Create using the DTM menu ( Civil > DTM )

Survey DTM- A TIN is created from survey data by triangulating the points and chains in the survey project. We can further control how the points are included in the TIN by setting up our SMD (Survey Management Database) features DTM control. This function allows us to control how a point or chain is stored in the TIN.

DTM - The DTM menu allows us to build TINs from a variety of sources. These would include :

Graphics
ASCII (txt)
DEM (Digital Elevation Model)
LIDAR (LIght Detection And Ranging)
Land XML

Within both the survey or the DTM tool there are tools and processes that allow us to to create the surface we need to move forward. Powerful editing and manipulation tools give us the ability to create surfaces where other programs will fall short.

The process for creating the basic TIN is a 2 step process:

1. Extract
2. Build Triangles

Unfortunately, life isn't always this simple. Graphics produced in DWG format may be locked or exploded. This would require additional steps to unlock, Join the exploded contours, or thin the number of vertices.

Further complicating this process could be crossing contours, or contours with no Z component. While we can work with this data it introduces more steps to the process.

Information on how to use these tools can be found in the Help (F1) menu inside PowerCivil along with "Watch It" links to videos on the tools. Additionally, there are basic training videos on http://www.powercivil.com/

Fig Above shows a surface created from SONAR Data. This data included over 15 million points. An algorythim disigned for LIDAR filtered the points maintaining accuracy to a more managable 1.6 million points.

Step Two - Create A Model

Up to this point getting our existing surface TIN created has been much of what we are used to. Many other programs create a TIN or DTM. In PowerCivil, we will now create a MODEL, a Dynamic, living breathing three dimensional surface of our proposed design.

Well, OK maybe I'm being a bit dramatic, it really doesn't breathe. The process is actually simple, we just changed some terminology. Before we outline how the Model is created and works, let's review the way we used to create that finished grade surface.

A sample workflow of other popular software:

•1. Create a DTM representing existing ground "EG"

•2. Place points at critical locations, these points are then assigned a Z component

•3. Triangulate points, specify side slopes

•4. Stitch (merge) the new surface into the EG surface to create a new surface

•5. Repeat this process adding subsequent surfaces

The problem with this method is we are unable to create a object using a relationship. Example: perhaps we need to keep handicap accessable parking at a specific slope from the front entrance of a commercial building. Additionally placement of points alone cannot model surfaces to the same degree of accuracy as lines and arcs.

Other problems arise when we need to make changes to this surface. These changes can require rebuilding of the entire surface for a minor change.

Site Modeler

As you see above the Finish Grade (FG) surface is usually a compilation of DTMs.

As my friend Michael Gilham (AKA: Mr. PowerCivil) once told me, "Site Modeler builds TINs". There it is, in a nutshell, the secret has finally been revealed.

Lets pull it apart a piece at a time to further unravel this Site Modeler magic.

1. First create a project (.GSF). The project will link to the DGN or DWG design.
a. We will need to establish the project preferences. These will allow it to display according to our company standards.
2. Create a Model by simply giving it a name.
a. A project can contain multiple models, each model will contain a minimum of one surface (object) but in theory can have an unlimited number of objects.
3. Objects (surface TINS) are created by horizontally locating CAD graphics then using the modeler "Add Elements" tools to assign a Z component to the graphics. There is no limit to the numbers of CAD graphics (elements) used in a object.

The good part is the entire process outlines in the 3 steps above has a Wizard to walk us right through it to the point we are ready to begin designing our new surfaces (Objects).

Dynamic Objects

You may be wondering, why not just call the objects DTMs like everyone else? What is all this talk of objects?

Now I wasn't in Jays head when he began naming things but here is where we are different.

The objects we create are basically free floating TINs. These objects tie to our base object (normally EG) using a slope we specify (3:1 by default). When two or more object surfaces intersect one another, the objects tie to each other and also to the EG surface. The order that they merge together is determined by a list FIFO (First In First Out)

Because these objects are in a site model project they remain "Dynamic". We can write the graphics representing the proposed grade at any given time and keep our proposed grading project in tact. This functionality leads to real productivity advantages allowing us to instantly update our site model.

Where Do We Go From Here?

Future posts will deal with and resolve issues from users both past and present. Stay tuned!

INTELLIGENT DESIGN

Mike Barkasi — Sun, 25 May 2008 06:09:00 GMT

When we set out to write this blog, the intent was not to make it another "CAD" blog, but to create a blog that will help teach or reinforce civil engineering principles that have been taught. This being said when we look at Civil Engineering we can see it is grouped into seven major divisions of engineering: Structural; Environmental; Geotechnical; Water Resources; Transportation; Construction; and Urban Planning. In practice, these are not always hard and fixed categories.

Seven engineering disciplines

A civil engineer working on a land development project could very well make use of all of the above mentioned divisions. Conversely an engineer working on a DOT (Department of transportation) project could also make use of all of these divisions. The question then is what separates the two? How are the projects different?

The difference lies in the breakdown of these divisions. A land development project may spend a greater amount of time in planning than a road project. While certainly the road project will receive its share of planning it will be subject to different constraints when compared to the Land development project.

Because the DOT or road project requires that a road be designed to move traffic as quickly and efficiently as possible from one point to another. Constraints on the project have to have some limits. Planning for these projects will usually be done at a higher level of government. While input from more local levels is given the need to maintain and develop this infrastructure is vital and local objections can be addressed with design features that do not alter the planned road layout.

An example of design features that we would consider secondary to the road way may be wall structures to limit noise from the road to adjoining communities. Access ramps or intersection development along the corridor could be added or removed. Retaining walls, storm water controls, lighting and landscaping are all part of the design that can be altered without gross changes to the planned road alignment.

Land Development

Before we begin comparing our Land development project we may want to decide what type of project we are looking at.

One local Ordinance defines land development as:

- The improvement of one (1) or more contiguous lots, or tracts or parcels of land for any purpose permitted in this Ordinance involving a group of two (2) or more buildings, whether proposed individually or cumulatively, or one (1) non-residential building on a lot or lots regardless of the number of occupants or tenure; or the division or allocation of land between or among two (2) or more existing or prospective occupants by means of, or for the purpose of, streets, common areas, leaseholds, building groups or other features.

From the definition above we could be describing a residential subdivision or a New School.

When we compare our Land development project, we will see the planning is quite different. While the planning is not entirely done at a local level, it is the local government agency that ultimately gives the final approval. This approval will usually require all other approvals such as DEP (Department of Environment Protection) permits and approvals first be obtained.

In addition to local Ordinances most local governments have planning comities to oversee all proposed development in the community.

As a civil designer on these projects the needs of the client or land owner take precedence in the design. Where our DOT project focused on getting people from point A to point B as quickly and efficiently as possible our site design will serve a purpose or invoke an emotion.

Typical Highway design will focus on moving people from point A to point B as quickly and efficiently as possible

The road traveling through a subdivision will often look entirely different from the highway or arterial road leading to the proposed development. Instead our site design road will be constrained by the topography, number of lots or parcels, Ordinance constraints such as prohibitive slopes, wetlands, riparian buffers, planning commissions and the designer's intent.

Once completed the road makes up a minor portion of the site design. Site grading and storm water control are major influences in site design. In the case of commercial site design we have different features or objects that we need to design for.

Parking lot design in itself requires the designer to be aware of the same storm water and grading as above but also the grading of the lot may be constrained by additional requirements such as ADA (Americans Disability Act) requirements, Local regulations governing slope of the lot or entrances onto the site, travel and turning requirements for the vehicles inside the parking lot, and more.

Because of these different requirements we need to ask ourselves if the design tools we use for Transportation and those for site design can be the same and effectively work in either scenario.

Low Impact commercial parking lot design is shown using Islands and plantings. Additional constraints in the form of accessibility, traffic flow, drainage and turning paths factor into the design

Because site design relies so heavily on relationships, the software we use should therefore also be able to create and maintain these relationships. The way we designed in the past did not allow for this "intelligent design".

PowerCivil breaks away from the previous mindset of CAD design into relationship modeling. The software had its beginnings eight years ago as "Site Modeler" developed by Jay Vose of Bentley Systems.

I believe Jay summed it up best in the quote below.

Here are my thoughts - they might be a little random but the seeds of site modeler were sown in the late eighties when I was a frustrated site designer. One thing to realize that since my first engineering job as a engineering tech while still in college I would write programs to help with tedious design concepts - detention ponds, water quality ponds, channels, culverts. The companies never approved of this since it wasn't billable time but I will be damned if I was going sit around and waste time redesigning all these things by hand. Anyway when I was doing one of my last site designs - a small grading, drainage and utility project the architect saw fit to move the buildings 18 times - reposition them, rotate them and changed the footprints. Each time it required me to re grade the site - redo the drainage and move the utilities around to follow the new layout. While doing this I said to myself that one day I would have to fix this - a few primary issues seemed most troublesome:

•1. I spend all this time and applying my engineering expertise to this design and all of its interrelationships and once input into the design packages it was all lost. Any change required reevaluation and recall of the design intent to insure I didn't violate some other required engineering constraint. Most everything was a series of relationships - horizontal and vertical and these are where my engineering experience was useful - not in redoing the design 18 times this was absurd!!!

•2. The plan content had to always be redone or at least checked every time I made a change - one forgotten elevation update and my boss would be buying the asphalt (this actually occurred - I had one old elevation in a pond detail once and it happened to be the one they staked it from so we had to rip the pond and adjacent parking lot out).

•3. Once could tell even back then we were building a 3 model - represented in more archaic fashion back then but it is what was going on. As I watched 3d design packages evolve there seem to be often more focus on the 3d aspect than the actual workflow. In reality site design is a series of horizontal and vertical relationships between features. The process is primarily done in this fashion - we horizontally locate features and then establish vertical relationships - we don't make the horizontal and vertical decisions at the same time so forcing me into a more 3d workflow was less intuitive. I want 3d when I want it. These thoughts led to the easily supported 2d workflow in site modeler and the when needed 3d. I didn't want to have a system that imposed a 3d workflow when it didn't fit my situation.

•4. The other aspect of available design packages, even now, is their reliance or primary focus on the linear and parallel aspect of design. Road products don't make good site products because in the site design world very few things are straight and parallel. Doing the math is easy when things are straight and parallel - that didn't help me as my stuff curves around is irregularly shaped and things like mathematical projections are in many cases impossible. Knots - projection errors and poor planar geometric considerations were the norm in packages that targeted road design - nobody attempted the hard solutions. While I didn't want to do the impossible - you ought to do a better job at solving these situations than the linear solutions provide. Almost nothing in site modeler depends on features be more linear and parallel - it was all targeted to provide the best reasonable solutions between very irregular feature paths.

•5. Design precedence - as I worked on these different design features I could see that some portions were more fixed than others (pavement sections versus slope tie downs) and that certainly the interfaces between them needed to almost float freely to as the backbones changed the interfaces would redo themselves automatically. This is where FIFO came from - determining the design precedence of object and allowing the interfaces to float and update in between them.