Geometry: Defining volumes passing through several others

Geometry
1

I use the ROOT geometry package to model certain components to be used by GEANT3, via the VMC interface. Having stumbled upon certain complications lately, I would like to ask you for opinions as well as possible suggestions.

The component in question is basically multi-walled pipe (represented by a series of tubes, some to describe walls proper and some to describe gasses between them) with two relatively complicated caps and flanges (represented by complex shapes) at both ends as well as several gas-circulation lines connected to different layers. The whole thing has been placed inside a two-walled box, sticking out on both sides. We want its GEANT description to be as accurate as possible because it is located close to the beam and can therefore generate many secondary particles.

Unfortunately even with complex shapes used to group volumes constituting respective components, it is still rather difficult to describe this object by a hierarchy of volumes. There are basically two problems I’ve run into:

  1. Placing end caps on the tube. With the gas volumes described by simple tubes, they either leave gaps or overlap with end-cap material. They cannot be made children of the end caps because there are two end-cap objects. On the other hand, end caps cannot be made children of gas volumes because in addition to having more than one of the latter - with different materials, too - it would require making them larger than they really are to avoid protrusions.

I see several possibilities of solving this, some of which sound to me better than others (but I’ll only say which later in order not to influence you):

  • add additional gas volumes which would accurately reflect the shape of the interior of end caps;
  • as above but converting the gas volumes into complex shapes incorporating these details instead of letting them remain separate volumes;
  • combine both end caps into a single complex shape.
  1. There are some volumes which pass through several others. Flow tubes start in internal gas volumes, pass through end caps exiting the pipe, pass through both walls of the outer box and only terminate in the world volume. Flanges are mounted in the walls of the outer box, protruding both outside the outer wall and inside the inner wall. The whole pipe itself only ends halfway between the two walls of the outer box, inside flanges. There are also support structures… I think you can see where this is going.

Here I have thought of the following possibilities, none of which I actually like much:

  • go against logical grouping and split each such volume into fragments, each of them only passing through to a single parent;
  • subtract the shapes of such volumes from parents (converting the latter into complex volumes in the process) to avoid overlaps;
  • bite the bullet, explicitly define such nodes as overlapping and try to set priorities so that correct materials are used by GEANT.

I shall be very much grateful for your suggestions on how to solve these problems properly!

2

Hi,

Thanks for the detailed post. I add in the loop @agheata , the maintainer of the Geometry package of ROOT: perhaps he has a high level suggestion to give.

For what concerns the more technical aspects, I think there is not much to say as this seems a more conceptual issue related to detector description rather than ROOT itself.

Cheers,
D

3

Looks like this answer comes very late, but for what it’s worth:

It is hard to give advice without having the object with in its current implementation before the eyes. However given your description, I would strongly advise against using either complicated booleans composition or making use of the overlapping feature. Since the beam pipe and surrounding structures are not sensitive, they are suitable to chop in multiple simpler volumes placed next to each other. Try to model the whole in sections along the beam axis, so that you can position the sections next to each other. Many experiments make a complicated Boolean solid describing this complexity, then make a hole to fit the pipe also as Boolean operation. All this is terrible for simulation performance, because as you say there are lots of particles tracked though outside of the pipe and Booleans are really slow…
So the rule of thumb is to chop in small volumes according to the complexity of the setup, represent holes also as placed gas volumes making sure that the setup is consistent when connecting this puzzle.