Step 1: Identify the two independent sources of redundancy.
In a triangulation network there are two separate types of condition (redundant) equations: (a) angle/shape conditions coming from the requirement that a triangle's angles sum to $180^\circ$, and (b) side/scale conditions coming from having more than one baseline to fix the scale.
Step 2: Compute the angle (shape) redundancy.
With $s = 8$ stations (6 triangles, as counted from the 18 interior angles), the minimum number of angles needed to fix the shape (not the size) of the network is $2s - 4 = 2(8) - 4 = 12$, the number of independent angles that fix a set of $s$ points up to a similarity transformation. Since 18 angles were actually observed, the shape (angle) redundancy is \[ r_1 = 18 - 12 = 6 \]
Step 3: Compute the scale (baseline) redundancy.
Only one baseline length is needed to fix the absolute scale of the network once its shape is known from the angles; the second baseline is then a check observation. With 2 baselines measured, the scale redundancy is \[ r_2 = 2 - 1 = 1 \]
Step 4: Add the two redundancies.
The total number of redundant observations is the sum of the shape redundancy and the scale redundancy, \[ r = r_1 + r_2 = 6 + 1 = 7 \] \[ \boxed{r = 7} \]