first draft for implementing link vehicle capacity counts handler #189

elizabethc-arup · 2022-03-07T01:18:10Z

This is linked to the link vehicle capacity feature request described in the issue #187, it adapts the link volume counts handler and adds vehicle capacities instead of counts so that an event like <event time="25201.0" type="entered link" vehicle="bus1" link="2-3" /> will result in the bus vehicle capacity for bus1 added to link 2-3.

For pt vehicles, the vehicle capacity is determined by mapping the vehicle ids to capacities using the functions in the TransitVehicles class in inputs.py and for cars, a default capacity value of 5 is used.

elara/elara/inputs.py

Lines 406 to 468 in 9a68ece

    
           class TransitVehicles(InputTool): 
        
               requirements = ['transit_vehicles_path'] 
        
               veh_type_mode_map = None 
        
               veh_type_capacity_map = None 
        
               types = None 
        
               veh_id_veh_type_map = None 
        
               transit_vehicle_counts = None 
        
               def build(self, resources: dict, write_path: Optional[str] = None): 
        
                   """ 
        
                   Transit vehicles object constructor. 
        
                   :param resources: dict, supplier resources 
        
                   :param write_path: Optional output path overwrite 
        
                   """ 
        
                   super().build(resources) 
        
                   path = resources['transit_vehicles_path'].path 
        
                   # Vehicle types to mode correspondence 
        
                   self.veh_type_mode_map = { 
        
                       "Rail": "train", 
        
                       "Suburban Railway": "suburban rail", 
        
                       "Underground Service": "subway", 
        
                       "Metro Service": "subway", 
        
                       "Bus": "bus", 
        
                       "Coach Service": "coach", 
        
                       "Tram": "tram", 
        
                       "Ferry": "ferry", 
        
                       "Gondola": "gondola" 
        
                   } 
        
                   self.logger.debug(f'veh type mode map = {self.veh_type_mode_map}') 
        
                   # Vehicle type to total capacity correspondence 
        
                   self.veh_type_capacity_map = dict( 
        
                       [ 
        
                           self.transform_veh_type_elem(elem) 
        
                           for elem in get_elems(path, "vehicleType") 
        
                       ] 
        
                   ) 
        
                   self.logger.debug(f'veh type capacity map = {self.veh_type_capacity_map}') 
        
                   # Vehicle ID to vehicle type correspondence 
        
                   self.logger.debug('Building veh id to veh type map') 
        
                   self.veh_id_veh_type_map = { 
        
                       elem.get("id"): elem.get("type") for elem in get_elems(path, "vehicle") 
        
                   } 
        
                   self.types, self.transit_vehicle_counts = count_values(self.veh_id_veh_type_map) 
        
                   self.logger.debug(f'veh types = {self.types}') 
        
               @staticmethod 
        
               def transform_veh_type_elem(elem): 
        
                   """ 
        
                   Extract the vehicle type and total capacity from a vehicleType XML element. 
        
                   :param elem: vehicleType XML element 
        
                   :return: (vehicle type, capacity) tuple 
        
                   """ 
        
                   strip_namespace(elem)  # TODO only needed for this input = janky 
        
                   ident = elem.xpath("@id")[0] 
        
                   seated_capacity = float(elem.xpath("capacity/seats/@persons")[0]) 
        
                   standing_capacity = float(elem.xpath("capacity/standingRoom/@persons")[0]) 
        
                   return ident, seated_capacity + standing_capacity

Currently, an if statement is used to deal with the transit_vehicle.xml version/legacy issue described in #188 as a temporary solution, it would be great if I could get some advice also on how best to resolve this. Thank you :)

fredshone

Looks good. Interesting concept.

Note that the transit capacities will be scaled in a pretty non-linear way, such that multiplying through by the scale factor might not make much sense. Nothing you can really do about this though.

If this becomes regularly used or project critical please add to readme and an example config.

btw, the post processing module could be used to automate further processing, eg flows/veh capacity.

Fixin

fredshone · 2022-03-07T11:54:34Z

elara/event_handlers.py

+            return veh_capacity
+        else:
+            return 5.0
+


veh_type = self.resources['transit_vehicles'].veh_id_veh_type_map.get(vehicle_id, None) return self.resources['transit_vehicles'].veh_type_capacity_map.get(veh_type, 5)

i would have written this as above, but on second thoughts, yours might actually be faster if it's predominantly cars so ignore.

Hi Fred, thank you so much for pointing this out as well as your comment regarding scaling. We had some further discussion on this and decided that car capacities are too heavily affected by scaling and won't be too relevant unless if something like ridesharing is investigated. We thought we would focus more on pt and pt crowding at this stage so I think I might add something very similar to the link passenger counts handler and make 'cars' an invalid mode.

elara/event_handlers.py

fredshone · 2022-03-07T13:14:02Z

elara/event_handlers.py

+        """
+        super().__init__(config=config, mode=mode, groupby_person_attribute=groupby_person_attribute, **kwargs)
+
+        self.groupby_person_attribute = groupby_person_attribute


i doubt this attribute groupby will ever be used? But also I think it doesn't really cause significant overhead so may aswell leave in

ah, I was thinking about this as well, I wasn't sure whether or not it should be kept in there. I will look into simplifying / removing it in the next draft :)

I think that there is not too much harm leaving it in, other than readability and maintainability.

I don't think there shoulld be any aweful consequences is removed other than a headache

fredshone · 2022-03-07T13:18:37Z

elara/inputs.py

+            standing_capacity = float(elem.xpath("capacity/standingRoom/@persons")[0])
+        else:
+            seated_capacity = float(elem.xpath("capacity/@seats")[0])  # new structure 
+            standing_capacity = float(elem.xpath("capacity/@standingRoomInPersons")[0])


this is fine for now, it doesn't get called many times so speed not a concern. I will keep an eye on this format to see if it is the new standard, but meanwhwile happy this supports both.

This is a result of a breaking change between versions 1 and 2 of the vehicles definition schema, as described in my comment on the issue raised by @elizabethc-arup.

We should check the entire schema and make sure Elara supports both versions, not just for this element, but for anything that has changed.

We should get some unit testing around the else clause - it's currently untested:

mfitz

The feature looks good and it's great to see unit tests.

There are a bunch of files in the diff that seem like maybe they should be in .gitignore - all of those binary files in tests/test_intermediate_data for example. I cannot see them being used anywhere - are they just an unimportant byproduct of running the tests? If so, we shouldn't be version controlling them. Similarly we have some files in tests/test_outputs and out/ - do these belong in the versioned repo? Some of them seem to have conflicts too: