Accessing Turns in Network Calculator

One of the main features of Release 7 is certainly the possibility to access turn attributes in the network calculator. After having added access to all transit attributes in Release 5, with this new development, the powerful tools available in module 2.41 are now available for all network elements and any combination of network attributes can be used in the calculations.

One major prerequisite for this development was a complete internal restructuring of the intersection table. Up to now, all turn related input data was stored sequentially in the intersection definition table. Only when performing an auto assignment, was the actual turn table constructed. Thus, the entire turn table was just a temporary structure, available only as long as the scenario contained a valid auto assignment.

Even though, from a user's point on view the functionality of module 2.31 is still the same (using the same input file structure and the same dialog, except for the dialog changes implied by the new possibility to display turn attributes as proportional bandwidth plots similar to those generated in module 6.14), the module has been completely rewritten to replace the intersection definition table by a permanent turn table structure. This restructuring has the following advantages for the user:

• Since the new permanent turn table is constructed immediately when the turn data is entered, it is now impossible to define more intersections than the data bank dimension ``turn table entries'' allows. Any attempt to exceed this limit immediately results in an error condition, so that consistency checks are no longer needed by the user to ensure that the auto assignment will not stop with a turn table overflow error.
• The turn table structure is adjusted automatically after any base network modification. Thus, it will always remain consistent with the current network, so that the user does not have to remove turn or intersection definitions after having removed the corresponding links or nodes from the base network.
• The turn user data items are now stored in their own internal file. Thus, their values can be modified freely, without worrying about the turn table size. (Remember that up to now each non-zero turn user data item occupied one entry in the intersection definition table.)
• Extra turn attributes are fully supported.
• The permanent turn structure allows full integration of turn data into the network calculator module 2.41 (see below) and the scattergram module 2.43.

When a data bank is used for the first time with Release 7, the old intersection definition table is automatically converted to the new permanent turn table structure (see the Release Notes for details).

Let us now turn our attention to the implementation of accessing turn data in the network calculator module 2.41. It can be summarized as follows:

• The new implementation of module 2.41 is fully upward compatible with the previous version. All previously available functionalities can still be used in exactly the same way as before. This implies in particular that no changes are needed to existing macros as a result of this development.
• The basic mechanism of the network calculator remains unchanged. All rules related to element and attribute type hierarchy, attribute inheritance, and result aggregation are still identical. As before, the type of an expression is determined by the minimum subtree in the network hierarchy which spans all attributes used in the expression.
• While all the underlying rules remain unchanged, three new attribute types have been added to the attribute type hierarchy:
  

  P: turn attributes
  M: outgoing link attributes
  K: k-node attributes
  

  See below for details of these new attribute types.

The attribute keywords for the new turn attributes are as follows:

Keyword: Attribute: Comments:

ptimau auto turning times after auto assignment only

pvolau auto volumes on turns after auto assignment only

pvolad additional volumes on turns after auto assignment only

tpf turn penalty function index integer between -1 and mfunc

up1 up2 up3 turn user data items

@xxxxx extra turn attributes as defined in module 2.42

tmpp1 ...tmpp9 temporary turn attributes lost when leaving 2.41

The keywords for the new attribute type M for the outgoing link are formed by appending the suffix ``m'' to the standard link keywords, i.e. lanm, lenm, timaum, typem, ulxm, vdfm, voladm, volaum and volaxm.

Similarly, the keywords for the K-node attributes are formed by substituting the ``i'' of the standard node keywords by the letter ``k'', i.e. k, fialik, inboak, ukx, xk and yk.

The following diagram shows the new attribute type hierarchy graph. Note that it is identical to the one used in Release 6, except for the insertion of the new attribute types P, M and K.

As can be seen from the above diagram, the turn attributes P define a new level just above the links and below the transit segments. From the point of view of a turn P, the incoming link corresponds to the standard link L and the outgoing link is accessed via the new link attribute type M. At the node level, in terms of the old turn nomenclature, J is the ``at-node'' (intersection node), I the ``from-node'', and the new node attribute type K defines the ``to-node'' of turn P.

In the hierarchy diagram, it is interesting to note that the turn level P is also connected to the transit segments. In fact, the transit segment S and the corresponding next segment N logically span a turn. This way, it becomes possible to use turn attributes in segment expressions (e.g. to copy auto turn delays to the corresponding bus travel times), or to aggregate transit segment data to the turn level (e.g. to compute turning volumes for transit vehicles or passengers). Also, since the new attribute types M and K are directly usable from segment expressions, they form a natural extension of accessing the attributes related to the next segment N, giving access to all link and node attributes which correspond to segment N.

We conclude this topic by mentioning just some of the many possible applications of this new development:

• Observed turn counts can be compared with assigned turning volumes (also by using scattergrams in module 2.43).
• Many intersection capacity and turn delay computations can be implemented directly within EMME/2, avoiding the use of external programs.
• Turn penalty functions can be coded automatically, based on street classification, intersection type and/or turning angles (e.g. using trigonometric functions and the xy-coordinates of the node I, J and K). E.g. the following expression yields the turning angle of turn P in degrees (0 straight through, <0 for left turns, >0 for right turns):

     180*atan(put((xk-xj)*(yj-yi)-(yk-yj)*(xj-xi))/put((xk-xj)*(xj-xi)
     +(yk-yj) *(yj-yi)))/3.1415926 +360*((get(1)>0)-.5)*(get(2)<0)

• Include auto turn delays in transit times of buses using the same turning movement.