|
GCM –
Development History
by Dejan GAVRAN
Even as a student of civil engineering, in early 1980-es, I was heavily
involved in computer programing. I think, at least quarter of the students
from my generation were pretty good programmers. It was a pioneering time and
our ambitions were far from the commercial ones. We were just enthusiastic
about making our students projects easier to do and, of course, about making
our examinations easier to pass.
In 1985-86, while preparing for final exam, I developed software
for drafting perspective views of the road from the driver’s eye position.
Unlike simple axonometry, characteristic for early releases of AutoCAD, the
software was drafting real perspective views on Commodore C64. For just one
colored wire frame picture there was 2 to 3 minutes of calculation. In
essence, the software was resolving many aspects of road design. The input
was pretty row: terrain surveys, basic elements of horizontal and vertical
alignment, elements of the typical cross section, criteria for applying
retaining walls etc. Before generating the picture itself, the software was
supposed to recalculate the entire roadway morfology.
In the years to come (1987-1990) I moved to Apple MacIntosh. I
continued with upgrading software for the perspective view of the roadway,
but also developed software for traffic network simulations (gravity model, Moore-Dial
algorithm, shortest path through the maze, traffic distribution etc.), for
finite element analyses of the pavement structures, for aircraft noise
simulation and mapping. By the end of the 80-es, I also did some software for
PC, mainly for the simulation of the aircraft movement. But, even when
working on Mac, which was exceptionally friendly for programming, most of the
programming code was not about the engineering calculations themselves, but
about mouse control, zooming in and out, plotting control etc.
In 1990. I definitely moved to AutoCAD on PC platform and it was
great relief for me. All general purpose options (mouse control, zooming,
plotting, general editing) were already present and the core of the
engineering problem could have been solved by using AutoLISP, very “fluent”
dialect of LISP (stands for LISt Processing), which happened to be one of the
oldest programming languages. The first serious undertaking in AutoCAD was
AeroCAD, developed in 1990-91 while working on master degree thesis, on
AutoCAD Rel.10 and AutoCAD Rel.11 platforms. It was entire package covering
all aspects of the airport’s airside design in plan projection: aircraft
movement simulation, design of runways and taxiways, apron and holding-bay
design, simulation of parking maneuvers, positioning of passenger bridges,
marking and lighting, positioning of navigational aids etc. The upper part of
vintage AutoCAD Rel.10 tablet (to be put on digitizer for easier and more
accurate mouse movement), containing AeroCAD commands, is given on Fig.2-01.
Fig.2-01. AeroCAD commands
Besides AutoLISP, AutoCAD Rel.11 (in 1990) enabled compiled C
code to run within AutoCAD. Thus, it finally became possible to seamlessly
incorporate the most time consuming calculations in AutoCAD based packages -
calculations like TIN (Triangulated Irregular Network) generation, prismoidal
volume calculations, extraction of profiles and cross sections, generation of
cut and fill slopes etc. While working on doctoral thesis (1993.), my
ambitions moved from 2D to 3D. In 1993. The first version of GCM was created,
and the name was AeroCAD 3D (Fig.2.-02). The intention was to create the
software for airport design in 3D, which is more complex than developing
software for road design only. While the road is typical linear facility, the
airside constitutes a system of interlinked linear and planar facilities.
Even the general cross section of the airside is more complex than the cross
section of the road, because it contains cross section of the runway, cross
sections of parallel taxiways and sometime aprons, and all these facilities
are interlinked with the complex and expensive elements of grading. General idea was to develop fully three-dimensional
software that would generate triangulated model of the airside. And all the
cross-sections and profiles would be simply cut out from the model’s
triangles. Also, the grading plans would be generated by cutting the model’s
triangles with the series of imaginary horizontal planes placed at the
constant intervals. Volumes would be calculated between the triangulated
networks as well. Finally, the finest shapes of grading were supported, even
the modeling of intersecting cut and fill slopes, which was out reach for
many software solutions of the era. Special set of commands for fine modeling
of runway-taxiway and road intersections was developed. The first version of
software was completed in autumn 1993, but for next four years it was not
commercially available. It was thoroughly tested and used on some in-house
projects.
Fig.2-02. AeroCAD 3D commands
After passing final doctoral examination by the end of 1996, I
was asked by the local engineering community to make the software
commercially available. Thus, in May 1997 (AutoCAD Rel.14 era), after
producing user manual and making some final touches, the software was sold to
first design bureaus in Israel and former Yugoslavia. The name of the
software was changed – it became GAVRAN – Civil Modeller, or GCM. Also, some
options that proved to be too specific for the airports were removed (such as
the options for checking the obstacle limitation surfaces, windrose analyses
etc.) and some new were added.
In order not to compromise GCM, I was refusing to sell the
product in larger quantities. Though there was a pretty good written manual,
I insisted on giving 4-5 day free course for each new customer. As I was
always a single developer, that fact was limiting sells to one or two new
customers per month. But, GCM was
quite a success. For example, it was possible only with GCM to model the
entire corridor of the motorway and then to extract the perfect cross
sections, containing not only the motorway itself, but also the service
roads, regulated riverbeds, tunnels
and bridges, even the nearby houses … everything that had been constructed by
using GCM’s triangulated models. By the end of the 90-es the software was
tested on very demanding tunneling projects in Austria and later in
Switzerland.
By year 1999, requests for GCM went higher. In order to satisfy
new customers, but still maintaining the high level of training, I decided to
create the first video tutorial. Thus, GCM2000 (AutoCAD 2000 era) came with
both written and video documentation (with more than 6 hours of video files).
GCM2000 introduced dynamic triangulated models, even the dynamic crossroads,
which was also ahead of the time.
GCM2004 upgraded dynamic 2D and 3D behavior in general and also
introduced dynamic behavior of simulated vehicle trajectories.
From technological and scientific point of view, GCM2006 was one
step back. Though full 3D had always been a main characteristic of GCM, local
engineering community was expecting simplification in the design of purely
linear facilities. Therefore, CSC macro language (Cross Section Constructor)
was developed. The language conceptually resembles graphical languages for
guiding the antique pen-plotters from 1970-90. era, but its syntax reflects
the geometrical reasoning specific to the cross section construction.
GCM2009 (or GCMx64) provides kind of a graphical substitute for
CSC coding. CSC code is now automatically generated from the graphically
defined road details. Also, GCM2009 introduces calculation of available sight
distances from the 3D model and the simulation of vehicle maneuvers is
performed by pushing and pulling the vehicle like toy.
GC2013 (or GCM++) further automates CSC concept, provides
support for speed analyses for both passenger cars and trucks, upgrades
optical analyses of the road, greatly improves 3D model dynamics by combining
the traditional 3D modeling techniques and CSC concept and introduces 3D
printing.
|