I don't believe that there are huge differences, however we have various controls under Project Settings - Computations - Surfaces or under surface properties that will do some different things to Bentley when it comes to what points we create between the surfaces when we compute a volume.
Typically when you do a Surface to Surface volume you take all the triangles from one surface and project the "nodes" into the other surface and vice versa so that you have the same number of points in the two surfaces and in exactly the same locations - then you also look at where breaklines of one surface cross the triangles in the other surface and vice versa and create additional points in the surfaces at those crossings. In this area we have settings that control how much of that happens (because e.g. if modeling point clouds where the data is super dense already you don't need to do that as it creates a huge increase in data that s not needed for accurate volumes.
Then you have boundaries - either ones you create or ones that we create because there isn't one present to constrain the model - there are settings in the Surface Properties that control that - Edge Length, Edge Angle being the most used ones and then there is the manual trimming of the edge that you can do - most people then put a Surface Edge Breakline into the model to constrain it so it doesn't keep reforming unwanted triangles. In addition to the outer edge breakline BC allows other boundaries to be added that act as Holes or Island boundaries - I don't know whether Bentley has that capability so that may be different.
We also look at crossing breaklines - if we find two breaklines that cross at different elevations in a surface, we will flag the instance to the user, but we will still form the surface model using the lower of the two breaklines at the crossing location and flag the higher of the two - again Bentley may do that differently
Vertical Tolerance - we have a vertical face tolerance of 1/10 mm but I believe Bentley's is 1/100 mm so they allegedly can handle faces slightly more vertical than we can in BC
Chording of curved lines and additional points along lines - we have tools that will automatically chord curved lines at an interval defined under Project settings to create additional points around Curves based on a Arc to Chord separation tolerance in order to form a TIN model from a curved line. That tolerance is user defined - I don't know how Bentley manages that - but if they have a setting you could match values and that should match the chording. Along long straight lines we also create extra points at intervals along the lines in the surface model - again that is a setting that if Bentley do something similar then you could match - If Bentley don't do that then you could set our value to a Large Number so it doesn't happen - again that may give a different tin model
Flat Triangles - when modeling contours or stockpiles on a flat stockyard, if you get a large bulge contour or an irregular stockpile boundary, very often you will get "Flat Triangles" that form across the Contour or between edge points of the stockpile - these are "flat Spots" that may not be what you want - the Flat Triangle Tolerance will try to force Triangles up a slope as opposed to allowing them to form across a flat area - this again could be different in Bentley
Lastly we also have the concept of an Alignment Based Surface that if initiated converts all the data in the surface to Station / Offset / Elevation coordinates. We then multiply (in background) the Station Values by 100 to stretch them out in the Long Axis of the alignment, we then triangluate the Station / Offset data as opposed to the XY data (this forces the TIN to form across sections in a "Road / Corridor" correctly i.e. between sections rather than using the traditional "Radial Sweep / Nearest Neighbour Delauney method of triangulation which can cause errors in corridors. We also have the Densification method that we taught Bentley how to do that will densify an Alignment Based surface based on an Arc to Chord tolerance setting that increases the number of "points" in the model through Horizontal Curves, Vertical Curves and through Transition Zones e.g. Superelevations - This modeling approach to my knowledge is unique and we will be different to others
Lastly there is the TIN Engine and its application of the Delauney Method - in the old days you seeded the model in one place and did the sweep algorhithm from one seed location. Today with Multi threading (on multi core PCs) you seed in multiple places and then sweep from many locations to make the modeling faster. How each system out there a) works and b) is constrained with settings will dictate the end result - this can create differences between the TIN that we create vs the TIN that Bentley may create from the same data. While the models should not be "significantly different" they will be different for all the above reasons.
Having said all this, if we are given a TIN model and import it as a TIN model we will form the exact same model (unless there are points at common locations where the points have different elevations or in some cases we see triangles that cross over each other - in which case in those areas we will flag the issue and correct it using the "Lowest Breakline / Flag High Breakline" method mentioned above. We treat every Triangle Side imported as a 3D Breakline and we make them internal breaklines so that the TIN is locked up and cannot be edited (because people tell us "that they want to use the Engineers TIN" but then complain that they cannot "Fix the TIN because it is broken in a few places" (we cannot win in that discussion ....)
When you have common data in two surfaces, we compute the difference in Elevation at each node between the two surfaces and between the Breaklines that cross triangle sides and form an Isopach which has a "Zero Datum" and all points are either Higher (where there is Cut from Surface 1 to Surface 2) or lower (where there is Fill between Surface 1 and 2). The isopach model is Triangulated and the volume is determined from the Isopach model.
Where we are supplied with a TIN model, the computations will be slower because every triangle side is a breakline in both models and the setting by default in BC is to "Track All Breaklines (in Project Settings". If you change that setting to Track No Breaklines then it will be faster to compute but still not as fast as a surface model created from Points and Lines.
We can also add data other than Points and Lines to our surface models - e.g. we can add a piece of elevation text, a 3D Block, we can add an entire surface to another surface and that I do not believe that Bentley can do so there may be some differences. We can also add these elements to a surface computed using Corridor Methods which I know Bentley cannot do.
Hope that this answers the question - what was the reason for your question.