Genie in a bottle
This is work related and as usual (as ever), standard disclaimer applies: That this opinion is entirely mine and neither does it reflect that of the company I work for, nor does it influence in anyway that I’m aware of.
Yesterday, we had a presentation by DnV of their upcoming new structural analysis software called GeniE which is part of the SESAM suite. They’re releasing it in the first or second week of September 2005. The presentation was a sort of heads-up on its capabilities and obviously the commercial bit.
I am a job engineer and I have no role in the commercial part of it. But here’s what I saw from the demo purely from an engineer’s point of view:
With GeniE, DnV intends to break away from the perception the industry has had about SESAM suite until now, which is, that it’s a raw file editing, command line and scripting intensive software. I think this change was inevitable and expected, given that software like SACS have virtually captured the market from what used to be SESAM’s domain eight years ago.
GUI: The interface looks better and more integrated. They have finally succumbed to OpenGL as a compromise for relieving system resources. Their rendering in (older) Preframe was better I think, but it was at the cost of performance. OpenGL is largely good, except for real-time rendering that still has some rough edges. For someone like me, who has used both the primitive method of inputting huge amount of model data by hand to fairly modern GUI wireframe interfaces, this is still a warm welcome.
Conventional (modern) method has been modelling wireframes using nodes and elements in their most prime order. GeniE takes it to the next level. In what they call it as a “concept modelling”, it adds a preprocessing functionality—a wrap to the wireframe modelling—by adding some nice bells and whistles, which go all the way to the most elementary level of modelling techniques, thus improving user experience. The end product is still a finite element model, but, the experience is a lot richer, better and enhanced.
Loading: GeniE takes away the pain of re-coding your loads when you change your finite element model in the slightest of ways. It automatically adjusts itself to the shifted members, elements or joints while allowing an exclusionary list, in which you can add IDs of elements or joints to be excluded. This is a fabulous feature—users will love this. Most users know how frustrating it is to re-calculate and re-distribute loads on these huge FEM computer models. This feature would come as a welcome relief and will be well received.
Speed: Most people—who have worked on SESAM earlier—know what it used to be in terms of computational time. We’d leave our computers overnight sometimes (we still do for wave response batch runs in SACS) so that the computations would run through the space frame model and produce results. Today’s computers have reduced that pain. It’s nice to know that GeniE is probably as fast as SACS—as I watched the demo yesterday. I wonder if SESAM has gone the way SACS did by intelligently trimming the modal partial differential equation to cut computational time without losing too-much of accuracy.
Report generation: It’s one of the best I’ve seen yet. GeniE produces a good XML file that can be read directly into Microsoft Offce 2003 for producing a detailed report. Ole Jan grieved that while everything looked good, he was still unhappy with the way the table of contents looked, and they were at best able to get some half cooked solution from Microsoft Support. So for now, he said, insert the table of contents command manually in Word. I wonder if DnV has done any testing on OpenOffice yet (Oops, I forgot to ask that!).
Suggestions for improvement: During the course of the presentation, I mentioned to the SESAM product manager, Ole Jan Nekstad that DnV software could add these following to further improve their already an impressive package, either before release or as an update later on. Ole Jan was appreciative of the suggestions and promised that they’d definitely take up some of them immediately and study others on the possibility of including in future.
- Modelling: To snap prismatic sections top-flush as in a realistic scenario such as connections of primary members to secondary members. At present, all structural analysis software, including GeniE do a centerline connection. This is generally fine, but miss the small moment out due to secondary beam offset in a rigid joint connection. Most engineers don’t take the trouble to offset secondary beams to flush with the top-of-steel (TOS), thus losing that end moment value.
- Multiple Undos from various sessions: Since SESAM maintains a record of every stroke that the user does in a plain text journal file, I thought it’d be a good idea not only using this for debugging, but also to add value to the software by adding a Undo session from various sessions and they need not be in a serial order. But I realize the difficulty as Ole Jan tried to explain that they were thinking of having a second database just for this, but would probably hinder performance. May be we wait for dual opterons to become a standard office desktop computer then :).
- Model exports: While GeniE has all the necessary import features from most industry used software, software vendors seems to be dragging their feet about providing export features. This is more than a surface itch; problems run deeper. Not everyone is willing to share and makes it easy for other vendors to help export; this is a widespread problem. We’ve seen this in everything from office suites, publishing software to the most advanced software in science and engineering today. Until people start using open standards, conversion is going to be a painful process. I was happy to learn that much of GeniE is based on open standards.
- Debugging: We have a major problem with regards to debugging. They’re never in plain english. Most users are stumped when they get an error code. Ole Jan said they’re looking to make GeniE respond better to error messages and make them more understandable.
- Analysis reports: To Ole Jan’s question on what should a typical output look like for a member or a string of members, I suggested it would be a good idea to add natural period for isolated members or for an isolated area. Until now, we’ve been modelling seperate free bodies to get this whenever we wanted to ascertain local effect of a heavy equipment sitting on a deck in operation trying to capture its dynamic properties.
- Outputs: Result outputs in XML and user control by stylesheets (such as XSLT) since they were already using XML to get a report output in Microsoft Word.
In all, GeniE looks like it’s ready to come out of the bottle with its best step forward.