Creating a Digital Twin Workflow with Speckle

Dear Speckle Community,

We are trying to create a digital Twin for a ship using IFC into Speckle, into Power BI where we can visualize the project and add data to the items.

We were facing some issues regarding max file size, which we solved with a dedicated Server.
And we are facing performance issues when refreshing the model in PowerBi.

Can someone help us setting up the correct viewer to solve this performance issue?
Perhaps we need to use Blender, Unity or Unreal as a Viewer?

How big is this IFC file? I think @dimitrie was looking to increase the default size on our public server.

Also, we’re releasing a new version of the PowerBI connector this week which should have some performance improvements too.
Finally, have you tried the viewer available in the new web abb?

It should be much faster!

1 Like

Hi @teocomi,

thanks for your reply.
Our IFC (from Autocad Architecture) is know approx 400Mb, but is increasing. We’re now at approx 15% of the project. At this time we are using Speckle on a dedicated 16 core 64Gb RAM Google server, this takes approx 12 sec to load our model (21sec on a 4 core 32Gb).

We are also looking at the possibillity using a local server, while we need the model to work without internet connection as well.

We are very interested in the updated Speckle viewer. Please let me know when it is available.
What in our opinion should help performance wise, is loading only the changes from Speckle to Power BI, instead of loading the complete model over & over again when refreshing.

Eager hearing from you.

1 Like

Hey @Sander_de_Boer ,

Our new viewer is already available at You can access the viewer package from here.

Yes, there are certain limitations put in place by Microsoft. In the meantime, i recommend trying the following options:

  • You can lighten your IFC model using a more flexible programming environment. We recently covered this in our Community Standup. Here’s a link to that:
  • If you are familiar with Power Query, you can use it to fetch only what you are interested in. This would speed up the receive times. Here’s a recording from our annual event, SpeckleCon, that covers this:
1 Like

Replace our workflow to Revit, Grashopper.
This is not really an option, we are now working with Cadmatic as our dedicated reversed engineering modelling tool. We prefer keeping it that way. The transformation from 3D DWG to IFC is done in Acad Architecture, which is increasing the model from 17 to 95 to 450 Mb. The main issue is the reloading in Speckle from the model (10 sec) and from Speckle to Power BI. (12 sec) with the dedicated Google Server.

Is it perhaps wise using another “viewer” as in Unity or Unreal, replacing Power BI?

See below our “workflow”

Power Query:
We tested this, but the Loading times did not decrease signifacantly unfortunately.

Hey @Sander_de_Boer ,

Have you tried opening .dwg in AutoCAD and directly sending from there?

1 Like