Auxiliary Models

The information below is valid for model commit #693038bf58447a6ff268b432b90b1afc4958a9b1 or earlier. For latest supplemental model documentation, please click this link.

Auxiliary models includes the following sections:
* Supplemental
* Trucks
* Update to Future Year

Supplemental

Supplemental models include external trips, group quarters and special generators. These trips are modeled in traditional 4-step trip model. External trips include internal-external, external-internal, and external-external trips. The first two types of external trip ends are estimated by trip generation model generation.py. The trip ends of externals-externals, internal-external, external-internal, group quarters and special generators are then balanced by generation.py, and distributed by distribution.py. OD matrices for supplemental passenger trips are then created.

More details can be found from PSRC 4K model documentation.

Below is a list of files residing in inputs\supplemental folder.

Name	Type	Description
generation	Directory	Inputs for supplemental trip generation
trips	Directory	Seed supplemental trips. These h5 files are copied to outputs/supplemental folder but then all files under this folder are deleted right before supplemental trip generation starts. The supplemental trip tables are actually generated from scratch. So these h5 files are not in use by BKRCast.
gravity_model.json	File	Gravity model coefficients by purpose, used in calculations of friction factors
inc_college_dict.json	File
inc_k12_dict.json	File
inc_size_workers_dict.json	File
inc_veh_dict.json	File
matrix_balancing_spec.json	File
mode_dict.json	File
purp_tod_dict.json	File
time_dict.json	File

Below is a list of files residing in inputs\supplemental\generation folder.

Name	Type	Description
external.csv	File	Productions and attractions for externals, being used in supplemental module (in new base year)
externals-2014.csv	File	backup of externals.csv used in 2014 base year model
pumshhxc_income-collegestudents.xlsx	File	PUMS household crossclassification data for college students
ensembles	Directory
landuse	Directory	land use for trip model (for EI/IE/EE)
pums	Directory	PUMS household cross classification
rates	Directory	trips rates for EI/IE/EE
specialgenerators	Directory	Not used

Below is a list of files residing in inputs\supplemental\generation\ensembles folder.

Name	Type	Description
puma00.ens

Below is a list of files residing in inputs\supplemental\generation\landuse folder.

Name	Type	Description
tazdata.in	File	PSRC 4K model land use, being used in supplemental module.
tazdata_2014.in	File	2014 PSRC 4K model land use, backup only
tazdata_2014.csv	File	2014 4K model land use in csv in spreadsheet format

Below is a list of files residing in inputs\supplemental\generation\pums folder.

Name	Type	Description
pumshhxc_income-collegestudents.in	File
pumshhxc_income-k12students.in	File
pumshhxc_income-size-workers.in	File
pumshhxc_income-size-workers-vehicles.in	File

Below is a list of files residing in inputs\supplemental\generation\rates folder.

Name	Type	Description
hh_triprates.in	File
nonhh_triprates.in	File

The seed supplemental trips are used when run_skims_and_paths_seed_trips is set to True in input_configuration.py. In that case, BKRCast used DaySim seed trips and seed supplemental trips to assign an initial demand on the network and create an initial set of skims. The seed supplemental trips can be updated from a previous model run outputs.

While transferring PSRC's SoundCast model to BKRCast, the seed supplemental trips were converted from PSRC's 4k zone system to BKR's zone system using psrc2bkr_trips.py under psrc_to_bkrcast_scripts. The conversion is no longer required as the supplemental trips from a previous BKRCast run (outputs/supplemental/[tod].h5) can be used as seed supplemental trips for any new BKRCast run.

Supplemental trips are generated only once and it does not change through feedback loop.

Generation (generation.py)

Trip generation model only estimate internal-external and external-internal trips. Production and attraction trip ends by purpose (all region) are calculated by multiplying the cross-classification table (tazdata.in) and trip rate tables (hh_triprates.in and nonhh_triprates.in). The following purposes are modeled:

• Home-based work (hbw)
• Home-based college (col)
• Home-based school (sch)
• Home-based shop (hsp)
• Home-based other (hbo)
• Non-home-based work (wko)
• Non-home-based other (oto)

tazdata.in

Column	Name	Description
101	PCTMF	Pct MF - Not used
102	HH1	HH - income1
103	HH2	HH - income2
104	HH3	HH - income3
105	HH4	HH - income4
106	GQI	Not used
107	GQN	Not used
108	POP	Population - Not used
109	RET1	Retail Employment
110	RET2	Not used
111	RET3	Not used
112	FIRES1	FIRES Employment
113	FIRES2	Not used
114	FIRES3	Not used
115	GOV1	Government Employment
116	GOV2	Not used
117	GOV3	Not used
118	EDU	Education Employment
119	WTCU	Not used
120	MANU	Manufacturing + WTCU Employment
121	FTEUNIV	College FTE
122	GQDORM	Group Quarters - College Dorms
123	GQMILH	Group Quarters - Military Housing
124	GQOTH	Group Quarters - Other

External Stations

There are 18 external stations are defined in BKRCast, same as in Soundcast. Below is the external locations. The handwritten numbers are BKR TAZ numbers used for external stations. The same table can be found in page 35 of PSRC 4K model documentation.

externals.csv

Column	Description	Values
taz	BKRCast TAZ where external stations are located
hbwpro	HBW production
colpro	Home based collection production
hsppro	Home based shopping production
hbopro	Home based others production
schopro	Home based school production
wkopro	Work based others production
otopro	Other based others production
empty1
hbwatt	Home based work attraction
colatt	Home based college attraction
hspatt	Home based shopping attraction
hboatt	Home based others attraction
schatt	Home based school attraction
wkoatt	Work based others attraction
otoatt	Other based others attraction
empty2
hw1pro
hw2pro
hw3pro
hw4pro
hw1att
hw2att
hw3att
hw4att

Special Generators

No special generators are defined inside BKR area. Three HBO special generators and one airport special generator are defined in externals. The same table can be found in page 34 of PSRC 4K model documentation.

The special generators are hard coded in emme_configuration.py.

Airport special generator annual growth rate is from the Port of Seattle. HBO special generators (Seattle Center, Exhibition Center, and Tacoma Dome) assume the same growth rate as auto externals. Truck special generators assume the same growth rate as employment growth which is the same as truck externals.

Airport special generator annual growth rate:
2014 - 2020: 3.65%
2020 - 2030: 2.69%
2030 - 2040: 2.18%

Airport trips are assumed 75% home based and 25% work based.

Group Quarters

Group quarters include college dormitories, military, and others(retired homes).
The outcome from the cross classification model is a production/attraction (PA) table, with 24 columns representing productions and attractions, and 1359 rows representing each individual TAZ. Special generators are then added to this PA table. The table is balanced to production (except col which is balanced to attraction). Some placeholders, like park_and_ride and dummy zones, are then added to make it complete TAZ system. The final products from generation.py are exported to prod_att.csv and gp_prod_att.csv in outputs folder.

Distribution (distribution.py)

Cost and distance between an OD pair are calculated by averaging AM and PM skims. Friction factors are then calculated based on the skims, friction coefficients, auto operating cost and value of time.

Name	Description
Fiction coefficients	gravity_model.json
autoop (auto operating cost)	16.75
avotda (VOT)	0.0303

After distribution, only external and special generator trips (daily trips) are retained and further split by mode and time of day. The final trip tables are exported to h5 files in outputs\supplemental folder.

Trucks

Truck model is also implemented in traditional four step model format. This inputs\trucks folder contains input data and settings for the truck model. It is borrowed from PSRC's 4K model and modified to fit BKRCast condition. For details about this truck model, read PSRC 4K model documentation.

AM auto network is also used as truck model network. truck_model.py is called to generate truck trips, distribute trips to appropriate TAZs, and create OD matrices in four time periods. Three truck classes are modeled: light truck, medium truck and heavy truck.

The socioeconomic data (tazdata.in, cross classification table) in the truck model is consistent with the auto model, except that the employment data are further split into more categories by multiplying MANU or WTCU jobs by factors. The employment categories used in the truck model are as follows.

Employment Category	Matrix	Split Factor	Factor Matrix	Notes
Agriculture/Forestry/Fishing	mo21(agffsh)	agshar.in	mo1 (agshar)	mo'manu' * mo'agshar'
Mining	mo22(mining)	minshar.in	mo2 (minshr)	mo'manu' * mo'minshr'
Construction	mo7(const)			stored inconst.in
Manufacturing (products)	mo23(manup)	prodshar.in	mo3 (prodsh)	mo'manu' * mo'prodshar'
Manufacturing (equipment)	mo24(manue)	equipshar.in	mo4 (eqshar)	mo'manu' * mo'eqshar'
Transportation/Communication/utilities	mo25(tcu)	tchshar.in	mo5 (tchshr)	mo'wtcu' * mo'tcushr'
Wholesale	mo25(whls)	whlsshar.in	mo6 (whlssh)	mo'wtcu' * mo'whlssh'
Retail	mo9(ret1)
FIRES	mo12(fires)
Education and government	mo29(govedu)

Truck productions are then calculated by

• Light: moltpro = 0.4102 * (moagffsh + momining + moconst) + 0.3466 * (momanup + momanue + motcu + mowhls) + 0.3281 * moretail + 0.1615 * (mofires + mogovedu) + 0.0928 * mohhlds
• Medium: momtpro = 0.0889 * (moagffsh + momining) + 0.0998 * moconst + 0.0858 * (momanup + momanue) + 0.2079 * motcu + 0.2552 * mowhls + 0.1637 * moretail + 0.0434 * mofires + 0.0297 * mogovedu + 0.0358 * mohhlds
• Heavy: mohtpro = 0.2366 * moagffsh + 0.3405 * momining + 0.0856 * moconst + 0.2661 * momanup + 0.0953 * momanue + 0.1075 * motcu + 0.1337 * mowhls + 0.0463 * moretail + 0.0000 * mofires + 0.0000 * mogovedu + 0.0000 * mohhlds

Truck attractions:

• Light: moltatt = 1.3161 * moagffsh + 47.9052 * momining + 0.2585 * moconst + 0.2581 * (momanup + momanue) + 0.5082 * motcu + 0.1464 * mowhls + 0.0501 * moretail + 0.1591 * mofires + 0.0966 * mogovedu + 0.1732 * mohhlds
• Medium: momtatt = 0.2831 * moagffsh + 14.8073 * momining + 0.0876 * moconst + 0.0538 * (momanup + momanue) + 0.0998 * motcu + 0.0352 * mowhls + 0.0123 * moretail + 0.0375 * mofires + 0.0160 * mogovedu + 0.0385 * mohhlds
• Heavy: mohtatt = motruck * 0.0988 * moagffsh + 5.0897 * momining + 0.0290 * moconst + 0.0208 * (momanup + momanue) + 0.0378 * motcu + 0.0087 * mowhls + 0.0032 * moretail + 0.000 * mofires + 0.000 * mogovedu + 0.000 * mohhlds

Land use restrictions for heavy trucks are applied to zones without allowable industrial parcels in both production and attraction. Truck special generators are then added to attraction. Both production and attraction are then further adjusted by a set of factors, which is defined in input_configuration.py.

Truck Type	Adjustment Factor
ltpro	0.544
mtpro	0.545
htpro	0.530
ltatt	0.749
mtatt	0.75
htatt	1.0

Attractions are balanced to productions. Impedance for three classes are calculated based on matrices from inputs/4k. Trip tables are balanced and converted to OD format.

Annual external medium and heavy truck trips are converted to daily and then added to medium and heavy truck trip tables respectively. Medium and heavy trucks are converted to passenger vehicles by:

• 1 medium = 1.5 passenger cars
• 1 heavy = 2 passenger cars

Time of day are applied to daily truck trip table.

Class	AM	MD	PM	NI
Light	0.194	0.346	0.240	0.220
Medium	0.208	0.417	0.204	0.171
Heavy	0.209	0.468	0.189	0.134

Final truck matrices are exported to auto.h5 in inputs/4k folder.

Update to a Future Year

Supplemental Trips

When creating a future year model, we should also update supplemental trips. Follow the following steps.

1. Update tazdata.in file and externals.csv file. Use the utility tool update_externals.py.
   Use the following annual growth rate for external stations (external.csv).
   2014 - 2025: 1.14%
   2025 - 2040: 0.77%

The aggregation from parcel file to 4K land use format (tazdata.in) is as follows.

• Retail = EMPFOO_P + EMPRET_P
• FIRES = EMPMED_P + EMPOFC_P + EMPSVC_P
• GOV = EMPGOV_P
• EDU = EMPEDU_P
• IND = EMPIND_P + EMPOTH_P

1. Change HORIZON_YEAR variable to appropriate year.
   All special generators in auto model will be automatically updated based on the following growth rates.
   Airport special generator annual growth rate:
   2014 - 2020: 3.65%
   2020 - 2030: 2.69%
   2030 - 2040: 2.18%

2. Group quarters are not updated at this time. We will update it when data becomes available.
   Future projection can be reached by using these annual growth factors.

Trucks

Follow the following steps to update the truck model.

1. Overwrite tazdata.in in truck folder with updated one from supplemental trip module.
2. Update truck externals with the following annual growth rates. Truck annual growth rate:
   2014 - 2025: 1.37%
   2025 - 2040: 1.48%
   These are external files:

File
medium_truck_ei.in
medium_truck_ie.in
medium_truck_ee.in
heavy_truck_ei.in
heavy_truck_ie.in
heavy_truck_ee.in

3. Update truck special generators using the following annual growth rates. These rates are the same as truck externals.
   Truck annual growth rate:
   2014 - 2025: 1.37%
   2025 - 2040: 1.48%
   These are truck special generators:

File
special_gen_light_trucks.in
special_gen_medium_trucks.in
special_gen_heavy_trucks.in

4. Other input files may need update as well, subject to data availability.
   These files are employment split factor files, truck time of day factors, gravity model coefficients, and truck operating cost.