How generative components could help to model the HVAC, Plumbing and Electrical services? We believe it is equivalent tool to dynamo (for revit). How advanced it is? How it can save modelling time?
Santhosh Kumar S said:How generative components could help to model the HVAC, Plumbing and Electrical services?
Yes, good question. I think the answer is 'not very well at the moment'... GC has 'Building Nodes' which are primarily architectural and structural nodes like Walls, Columns, Windows / Doors etc.... not HVAC nodes like Ducts, Fans etc or Plumbing nodes like Pipes, Valves etc or Electrical nodes like Cables, Light fittings etc. Without these, there won't be much 'exploration of variants' or flexing of the model based on parametric manipulation.
Santhosh Kumar S said:We believe it is equivalent tool to dynamo (for revit)
Kind of. I think that Dynamo is much more integrated with R'vit. The reason for this is that R*vit was set up to work centrally as a database. A change or 'revision' anywhere is propagated and managed centrally by Rev*t. In comparison, each tool in Aecosim is responsible for managing the propagation of its own changes, old skool CAD-development style. There does not seem to be much in the way of commonality or API which will make it difficult for the GC or Aecosim team themselves to write GC wrappers / nodes for those MEP objects mentioned above.
"How advanced it is?"
GC? I think quite advanced when it comes to geometry problems. It also has a Genetic Algorithm extension, which is also cloud-based and allows for large scale batch processing. It is also based on dotNET which should allow it to leverage Mstn, without having to grapple with unmanaged native code. OTOH, the C# API support in Mstn CE has taken a long time to emerge. Verticals like Aecosim are even slower, being based on old VBA and COM API's which are being re-written for 64bit etc. Calling these functions from GC is not as straight forward compared to Dy'mo. Although, this may change sooner rather than later... who knows?
Santhosh Kumar S said:ow it can save modelling time?
The term 'generative' in GC does not have the same meaning as 'generative design' has currently, which favours the 'bio-mimicry' origins of the term: Genetic (and other) algorithms that iteratively traverse large solution sets looking for optimal solutions. I think that GC's 'generative' toolset and moniker owe more to Catia's Generative Shape Design workbench than any Genetic algo's.
So, how can GC save time when modeling... MEP designs? This is a very pertinent question, especially when you realise that using GC (like almost any productivity tool) will require a substantial investment in learning and resultant down time. It would be good to understand what GC/ABD has on the road map to avoid investing in something that will run out of gas, leaving you isolated because the software vendor can't justify the cost due to weak uptake. A Chicken-and-egg question that Bentley seems to like to approach in collaboration with big Asian engineering houses who have substantial coding resources in house, these days.
1. Inputs: Most MEP designs rely heavily on inputs from the environmental and envelope model, typically provided by the architect and/or other engineers. The number of occupants accommodated in a building will drive the number of WC's which will in turn drive the water and drainage system sizing. The geometry and facade design will drive the heat gain in the various rooms and in turn drive the HVAC design. So, the architect's GC model can hopefully be used to flex and explore different options to chase down the most efficient MEP solution. The MEP nitty gritty model elements (ducts, pipes etc) do not need to be GC nodes -in the first instance- to benefit indirectly from GC.
This however assumes that the MEP user can use GC's Scenario Services to automatically calculate the MEP impacts of the iterations produced by flexing the architecural GC model. STAAD seems to be hooked up, Not sure about AES or Hevacomp.
2. Routing: A big part, time-wise, of any MEP design is routing. GenMEP has attracted some attention with its rules-based smart routing / clash avoidance tool. I suppose it is 'generative' and might be using a GA or it could be something rules-based like Bentley's PlantWise or Selvaag Bluethink which been doing auto-routing for decades. Really applicable in the early 'form finding' stages of the design where GC is typically used. In any case, auto-routing needs a different computational approach to what GC does, which is more 'scripting'-based.
Maybe, we will see PlantWise become a Companion app in Aecosim or BRCM at some point.
3. 'Fine-grain' automation: GC would be good at providing a visual programming tool to support what I would describe as 'fine-grain' ad hoc automation tools. This example from the Catia world is based on automatically placing parts based on the selected pipe sizes / materials. This kind of rules-based 'if-then' branching type processing can also be scaled up using 'templates' to create whole systems. Yea.. there will be places where the results would need to be topped and tailed, but it will still be a lot quicker that doing things manually. Good for early stage costing? Cost is a key cumulative dimension to BIM that will have huge spillover impacts on the whole BIM exercise.
Like Dy'mo, this kind of automation can also be extended to automate the drawing production and quantification workflows. The big advantage tools like Catia's KnowledgeWare, Dy'mo and in theory GC have is that they provide 3d geometry-savvy scripting support, where you the domain expert would otherwise have to 'roll your own' using VBA or LISP or whatever, which is a pretty big blocker. And, let's face it, even if you didn't have to deal with 3d issues, having a visual programming interface is a lot more productive for a lot of the modest-but-legion (realistically only be done by your (super) user, not your expensive non-domain expert pro-coder) automation wins that will add up. Notice how Dynamo was able to extract 'database' info like pressure drops from the R'vit MEP model without any coding.
Going beyond modeling, this kind of Catia-style, wall-to-wall support for scripting/automation would also enable direct-to-fabrication and plug into all that engineer-to-order goodness that the manufacturing world offers.
Hi Santhosh,
As I pointed out in my initial response and as Dominic confirmed in his responses, GenerativeComponents covers architectural and structural disciplines.
GenerativeComponents does not cover MEP.
MEP modeling with GenerativeComponents can be approximated by using other attributed geometry; however, that would be a workaround and could lead later to other pitfalls (e.g. when attempting to get good IFC information from such data), as one of my colleagues pointed out to me in a conversation about your inquiry.
Therefore, no training materials for modeling MEP with GenerativeComponents has been generated. Consequently, using the work around approach is left to the user's ingenuity.
I hope this suffices to conclude this discussion.
Best regards,
Volker
Answer Verified By: Santhosh Kumar S
Please let me know, In future, if Bentley had an idea to facilitate the MEP stuffs in Generative Components.
Your name and e-mail address is now on file with the corresponding enhancements in GenerativeComponents' backlog.
Regards,
Mueller, Thanks for your Answer for Santhosh Question it helped me a lot to understand.
I have one more Question, Can I perform all my design in R*Vit - Dynamo and can able to transfer all data to Aecosim. Is this Possible without loss of data transfer.
Hi Jagannatha,
Jagannatha Reddy said:without loss of data transfer
Short answer: no.
Longer answer: Revit with Dynamo on one side and AECOsim Building Designer (OpenBuildings Designer) with GenerativeComponents on the other side use their own proprietary representations. Therefore, any transfer of those proprietary data will have to flow through some non-proprietary compromise format, for example IFC. IFC does not carry parametric information.
If your result is intended to be fully functional in OpenBuildings Designer with GenerativeComponents, then it is most effective to start your work there.
Best success,