I have found an issue with Auto adjustment when used to connect an underground network to several entrances. Although the adjustment and connections seems to work nicely, there is an issue after a network is set to be connected to multiple entrances as some of the branches get disconnected from the rest of the network. This happens when Adjustment is set to "None" and this shouldn't happen as it remain important to be able to assess the network with or without adjustment.

As explain before, to keep things as simple as possible in setting up connections, I consider the network as one and only connects the extra-entrance to this one network. In this case, I downloaded the entire data set with the network already built during acquisition phase in TopoDroid, so the data imported in CRP are chronically ordered in the table: first data acquired at the top, last at the bottom. All the survey shots/branches are connected as one network and to the "main" cave/entrance. I pretty much only added the reference point of the “main” cave (2001/0.0 – 2001/0.0). In this case, the main network was checked, consistent and correctly connected. Then I create each extra cave for each "entrance", filling in coordinates and altitude and added a connecting reference points both in the main cave and the entrance caves. When the network was fully set and ready for Auto adjustment but before Auto adjustment was ever applied, I could see that some of the branches got disconnected from the rest of the network. However, when I enabled Adjustement "Auto", everything connected OK: Entrances are all connected and survey shots adjusted. When I disable the Auto adjustment again, then the disconnections happen again.

Looking closely to all the branches, it looked like the acquisition direction of the branches could have been the reason for such an issue. By acquisition direction I mean if the branch was acquired going away from the main entrance and toward the extra entrance or going toward the main entrance from the extra entrance. However after a long investigation, the issue doesn’t seems to be related the acquisition direction and it seems to be related to the ordering of the survey shots in the table.
It’s a quarry with 2 entrances (0.0 and 1.0) and 3 vertical air ducts to the surface with known position (2002/3.0, 2002/4.0 and 2003/5.16).
Stations 2001/0.7, 2001/1.6 and 2001/1.3 get disconnected from the network when connections to extra cave entrances are set and Auto-adjustment is disabled.

It took me some times to identify the root cause as there are many specificities that could have contributed to this issue:

Branch 1.x
- Acquisition direction toward the main entrance
- Original survey shot 2001/1.2 > 2001/1.3 as a large angular error as station 2001/1.2 was too close from a metal frame in the gallery causing a magnetic disturbance. So we had to redo some measurements on a second session so there are a couple of disabled shots from first “series” (2001) and a couple of by-pass shots from a second ”series” (2002) with a new station 2002/1.2b.
- Has an internal loop: branch 2.x

Branch 2.x
- Internal loop
- Not connected to any extra entrance

Branch 3.x:
- Acquisition direction toward the main entrance.
- First shot is reversed in a vertical air duct
- Has an extra survey shot after the connection node

Branch 4.x:
- Acquisition direction toward the main entrance.
- First shot is reversed in a vertical air duct
- Has an extra survey shot after the connection node
- Same connecting node as branch to 5.16

Branch 5.x:
- Acquisition direction going away from the main entrance.
- Same connecting node as branch from 4.0


And here is what I have done/checked to narrow down the issue:

I have checked the network connectivity again by removing all the connecting points and extra entrances and I really don't see any issues there.

I have checked each branch connection individually with and without auto-adjustment:
>>> Branch 1.x gets disconnected at 2001/1.3 (area were we had to reshoot and bypass original station)
>>> Branch 2.x is OK in auto adjustment and in loop mode.
>>> Branch 3.x gets disconnected at 2001/1.6 (connection node)
>>> Branch 4.x gets disconnected at 2001/0.7 (connection node)
>>> Branch 5.x is OK

I have investigated the strange disconnection on branch 1.x at 2001/1.3: the by-passing survey shots via 2002/1.2b were acquire on day 2 and therfore placed further down in the original table, so below the 2001 series (day1). This caused no issue in connecting the branch 1.x to the rest of the network before any adjustment/connection settings were made but it causes an issue after connection to extra-entrance settings were made. If I move the by-passing survey shots to/from 2002/1.2b between lines containing 2001/1.1 and 2001/1.3, then the disconnection issue is resolved at 2001/1.3 but moves down the branch to the connection node at 2001/0.6.

I have tested some dual branches combination to understand why the branch 5.x, which is OK on its own, gets disconnected when all the entrances are connected. And the branch 5.x gets disconnected only if the branch 4.x has its connection to its extra entrance set. Both branches share the same connection node to the rest of the network and I suppose branch 4.x is prioritized over 5.x due to order in the data table, get disconnected first and then 5.x disconnection is just a collateral damage.

I have confirmed that internal loop (branch 2.x) causes no specific issues.

I have confirmed that the extra shots beyond the connection nodes (2001/1.6 and 2001/0.7) are not causing the issue. I removed them and there’s no changes.

I have inverted the survey shot in the air ducts (4.1>4.0 >> 4.0>4.1 and 3.1>3.0 >> 3.0>3.0) to ensure all survey shots were in the same direction for the entire branch but this didn’t solved the issue.

I have also tried to invert each survey shot to ensure the shot direction was flowing away from the entrance, but this did not solved the issue. This was done without rearranging the line order in the table, just by editing azimut (+/-180°) and inclinaison (+/-).

Least but not last, I have rearranged all the original survey shots (not inverted) for the problematic branches, moving the connecting node at the top and flowing down the table to the extra entrance… and then surprise! Surprise! Everything became good, both in “Auto” and “None”…

What a journey! It finally comes to something "simple", but it wasn't easy (to me) to find the root cause in this issue and I hope my explanation are a bit clearer than mud.

I don't think it is normal that some branches get disconnected from the “main” cave when no adjustment is applied: primarily because they belong to the “main” cave and are correctly connected to it but also because it becomes complicated to assess the difference before and after adjustment. I strongly believe it shouldn’t be necessary to reorganise the data line in the table for CRP to handle the set of data in a correct way making all this a complicated experience to have it right. All the necessary details are in the data table for CRP to connect the survey shots together correctly without being dependant of the ordering in the table. The fact that the network connections are correct before the extra entrances were added just proves this is possible. I hope to see this issue resolved to keep thing as simple as possible.

I have attached the files for you to have a look with original ordering and issue... and the version "fixed".