PARAMICS Plugin
Contents
1. 18 PATH ATMS Center SWARM Ramp metering control Plugin 4 Technical Supports 4 1 Limitations of this plugin 1 Failure management SWARM in the real world considers the situation that loop detector data are missing or deficient However in the simulation world detector data are obtained based on another plugin module loop data aggregator which provides aggregated detector data as accurately as possible As a result this developed SWARM plugin works with good loop detector data all the time The failure management part of SWARM never has chance to be used 2 Queuing strategies Based on the design document of SWARM the central SWARM system will have four options about queuing control 1 Use local controller strategy i e when a queue is detected go to maximum rate 2 Gradually increase rate to maximum 3 Ignore queuing for a fixed number of periods then switch to gradual rate increase 4 Totally ignore queuing However the source code shows that there are only two options using the local strategy and disable queue strategy See the following for reference typedef enum RMS queue strat QUEUE STRAT UNKNOWN 0 DISABLED QUEUE STRATEGY LOCAL CONTROLLER RMS queue strat enum In the Traffic Engineer s manual of District 7 there is another description of queuing strategy Local queuing strategies such as forcing maximum rates when a queue is detected affect SWARM 1 operations The final SWARM 1 metering rat2. 3 swarm control containing the initial setup of SWARM to the target network Unlike the parser system of PARAMICS which allows flexible grammars the formats of these input files are rigid and thus any problem may cause that the SWARM plugin cannot be loaded As a result we hope users can use the example input files we provide with this plugin as the starting point to make your own input files for avoiding editing problems In order to correctly use this plugin in a target network the infrastructure layout of the target freeway network needs to be obtained from the proper government agency such as Caltrans The infrastructure layout includes the number of lanes and locations of loop detector stations This layout is also the basic information required for network coding in PARAMICS An example of this infrastructure layout can be found in APPENDIX 1 3 1 Adding detectors 3 1 1 Real world ramp metering system The configuration of a typical ramp metering system in California is shown in Figure 3 Mainline Station e _ Passage Demand Queue Figure 3 Typical ramp metering configuration PATH ATMS Center SWARM Ramp metering control Plugin Five types detectors can be possibly installed for a ramp meter including on ramp detector demand detector passage detector queue detector and ramp HOV detector The on ramp detector is used for counting total number of vehicles entering freeway from entrance ramps The demand
3. u eee ett bends shaven ON OY RE VIRI a a i 17 3 8 Fine tuning parameters of SWARM eeeeeeeeeeeeeee eee eneeennnetnnnt ee nnnt enn 17 4 Technical SUP POM Saw dore aeo n a ation t inde T 19 4 1 Limitations ob this DIUEITI udo e oie Dt gebat E E bens ORIS 19 4 2 Contact THEOFERQE DE go n be RA ims eM o Ud LC Ue dE 20 4 3 uico cM C m 20 APPENDIX 1 An example of the infrastructure layout obtained from Caltrans 21 PATH ATMS Center SWARM Ramp metering control Plugin 1 Introduction The purpose of this plugin is to allow users to implement the System Wide Adaptive Ramp Metering System SWARM of Caltrans in the PARAMICS simulation environment This plugin is developed based on source codes of SWARM obtained from Caltrans PATH ATMS Center SWARM Ramp metering control Plugin 2 Plugin implementation 2 1 Algorithm description SWARM is a coordinated traffic responsive ramp metering operation strategy which is developed as a component of the Advanced Transportation Management System ATMS at Traffic Management Center TMC by National Engineering Technologies NET System It has drawn wide interests because of the introduction of traffic flow forecasting in the algorithm The initial field tests were attempted in the Field Operational Test FOT of an integrated corridor level adaptive control system from fall 1994 through spring 1999 in the City of Irvine California The system
4. SWARM 2a algorithm 1 0 Target speed reduction 0 1 20 0 range 0 1 to 20 0 precision 1 The sixth part includes parameters of the SWARM 2b algorithm 0 1 Smoothing factor for the final storage computation 0 0 1 0 range 0 0 to 1 0 precision 2 0 85 Fraction of saturation density equivalent to LOS D 0 0 1 0 range 0 0 to 1 0 precision 2 The seventh part includes parameters of Phase 2 failure management used to decide if the detector data at a detector station can be used in SWARM 0 66 Fraction of good lanes required for station statistics 0 0 1 0 range 0 0 to 1 0 precision 2 The eighth part includes parameters of the startup and shutdown strategy 6 Start metering counter 1 20 range 1 to 20 precision 0 20 Stop metering counter 1 30 range 1 to 30 precision 0 11 PATH ATMS Center SWARM Ramp metering control Plugin The parameter in the last part is not a parameter of the reat world SWARM but a parameter of the SWARM in the simulation world Using speed estimation 0 or PARAMICS speed 1 0 1 range 0 to 1 precision 0 These parameters are universal parameters used in SWARM If you want to know more about the functionality of any a parameter in SWARM please find related information from the references listed at the end of this document 3 3 Preparation of vds control vds control includes the information of loop detector stations in the simulation network The format of this file is number of
5. freeways 1 polling cycle 30 number of mainline detectors 9 405n5 74ml 405 N N 5 74 5 45 yes 405n5 55ml 405 N N 5 55 4 45 no number of off ramp detectors 5 405n5 55fr 405 N N 5 55 There are three parts in vds control The first part includes the general information i e the number of freeways and the loop detector aggregation cycle i e polling cycle polling cycle should be the same as report cycle in the loop control file and metering rate update interval in the swarm control file The format for mainline detectors is Freeway ID direction pri direction loop name post mile number of lanes saturation density whether this loop station is a bottleneck The format for off ramp detectors is The VDSs listed in the vds control file are ordered by freeway id direction from downstream to upstream direction can be one of N S W and E pridirection refers to the direction of post mile and thus can also be one of N S W and E For example if the post mile at downstream is higher than that at upstream pridirection is the same as 39 66 direction saturation density refers to the density of the loop detector station at 12 PATH ATMS Center SWARM Ramp metering control Plugin capacity It can be a value between 40 45 The SWARM plugin provides capabilities to calculate the saturation density of mainline detector station based on simulation results and report to an outpu
6. is currently tested in the I 210 freeway within the Los Angeles transportation network The SWARM algorithm actually consists of two independent algorithms SWARM 1 and SWARM 2 SWARM 1 is a forecasting and system wide apportioning algorithm us ing a forecasting methodology SWARM 2 includes two local traffic responsive ramp metering algorithms SWARM 2a and SWARM 2b We provide a brief description of these three algorithms below Details about them can be found in the listed references 2 1 1 SWARM 1 SWARM 1 forecasts the traffic state at predetermined problem points bottlenecks and adjusts metering rates based on forecasts It treats the freeway network as sections Each section is defined as the two adjacent detectors that have reliable data outputs The operation of SWARM 1 is based on traffic density with the goal of maintaining realtime density below a pre determined saturation density for each section of freeway The linear regression and a Kalman filtering process are applied to pass detector data to forecast a density trend at each detector location for each control interval The time into the future to forecast is a tunable parameter named Tcrit Once the forecast density trend is obtained it can be combined with Tcrit to calculate the excess density the portion above the saturation density in the following Figure 1 Excess density is used to calculate the target density for the next metering cycle Target Density Current De
7. PATH ATMS Center SWARM Ramp metering control Plugin PARAMICS Plugin Document SWARM Ramp metering control Lianyu Chu Henry X Liu PATH ATMS Center University of California Irvine Plugin Compatibility V4 Release date 3 20 2003 522 Social Science Tower Irvine CA 92697 3600 URL http www its uci edu NIA H CALIF Sey 08 L mf L1 University 6f California irvine P PATH ATMS Center SWARM Ramp metering control Plugin Table of Contents Table OF COMeNtS Ae Sais eue aed OEE d eu E 2 Ls introductio cerdo d aee tae dais iate o ee 2 2 PIU SiN implementati eee HEY AIAX AERR E ES EE AU POP ERR S UR GE TERUUUS 4 23 LA IeortiEinn description oeste tomos totes usato Rc bNosed a ane ase breed 4 2 2 Development framework iui ee mene RSS xa He rag EH rie ee ades et Ue eaa re vertu UIS 5 2 3 Pseudo CODES osi enn ecd e aud Dette ii iet cm ee tiia 6 3 Step by step user mianual erase eret en nA NR ENS EE MIR TR NS sbseeeeaavebsadesndeneonseavedadeons 9 Dil Adding detectors i eee ei n CO EU RED NERIS ERUIT ARE RAI SUE NEN aa 9 3 2 Preparation of swarm global 4 ict tren aer terea to tede Re E deae 10 2 3 Preparation Of vlsccontEol scootao oh pite ieb eme edup dh E gato ded mss 12 3 4 Preparation of swarm control eseeesesesseeeeeeeeeee nennen eene 13 3 5 Loading Pla gir e ER 14 3 0 COMUNE Te S ose epos vei oles Nae eM MC EO eS 15 3 4 Brtor CHECKING
8. and passage detectors i e corresponding to the check in and check out detectors are used for the operation of on ramp signals The demand detector employs to initiate green and the passage detector employs to return the signal to red The queue detector is located at the upstream end of the entrance ramp used for detecting the excessive queue length in order to avoid interference with the arterial traffic The ramp HOV detector is used for counting the number of carpool vehicles entering freeway from entrance ramps Caltrans currently has three ramp metering systems SATMS SDRMS and SJRMS TOS All of them use the 170 type controllers as hardware SATMS SDRMS and SJRMS TOS are names of their software algorithms installed in the 170 controllers The SWARM algorithm is designed for the ramp metering system of SATMS used in District 7 and 12 The local controller and its ramp metering software of SATMS support centralized metering control i e the application of the requested metering rate from TMC 3 1 2 Detectors required for this plugin The demand detector has been used in the ramp metering control plugin which is a supporting module of the SWARM plugin The ramp metering control plugin acts like the local controller in the real world In order to make this plugin work three additional detectors or detector stations are required to be put to the simulation network They are the on ramp detector queue detector and the mainline de
9. e can be one of has the following DISABLED MODE SWARM 1 SWARM 2A SWARM 2B SWARM 1 2B SWARM 2A 2B SWARM 1 2A SWARM 1 2A 2B LOCAL TOD and LOCAL LMR 13 PATH ATMS Center SWARM Ramp metering control Plugin All ramps should be included in swarm control If it is a freeway to freeway or unmetered ramp the metering mode can be set as DISABLED MODE minimum rate control is one of the following two tTOD TABLE MIN ABS MIN default rate control is one of TOD TABLE DEFAULT and ABS MAX swarm startup strategy is one of RUN SWARM ANYTIME and RUN SWARM DURING TOD ONLY 3 5 Loading plugin The names of this plugin files are swarm dll Modeller Plugin swarnrp dll Processor Plugin The SWARM plugin depends on other two plugins ramp metering control and loop data aggregator These two plugins should be specified earlier than this plugin in the plugins or programming file i e loop agg dll ramp controller dll swarm dll If you want to implement SWARM with a queue override strategy you will need to disable the queue override strategy in the SWARM plugin through specifying queue detector as N A in swarm control Then you need to add on ramp queue control plugin to the plugins or programming file with the following sequence loop agg dll ramp controller dll queue control dll swarm dll In addition in order to correctly load and run this plugin please satisfy t
10. e developed the SWARM plugin with the following pseudo codes Loop data polling time i e the time to update metering rate Calculate loop station statistics For mainline VDS Get aggregated loop data from the loop aggregator plugin Calculate normalized station density occupancy speed Calculate the average vehicle length For ramp VDS Get aggregated loop data from loop aggregator plugin Query the current time of day metering rate from ramp metering plugin Query the current default rate control Query the current minimum rate control Check the status of loop detector station at known bottlenecks If bad search for a good upstream loop detector station to replace it Calculate update saturation density PATH ATMS Center SWARM Ramp metering control Plugin For each mainline VDS Collect enough VDS data in one day If enough VDS data SAMPLE_SIZE_SAT_DENSITY Calculate the saturation density for the VDS SWARM 2a For each on ramp Starting from the furthest downstream ramp on the freeway Calculate station speed at upstream loop station ft sec Calculate time headway from upstream loop to downstream loop Calculate the speed reduction per mile Calculate the desired metering rate Minimum rate control Maximum rate control Rate smoothing Startup shutdown strategy SWARM 2b For each on ramp on freeway Starting from the furthest downstream ramp on the freeway Calculate estimated storage i
11. e will have a lower bound equal to the rate selected by the queuing strategy The difference between the forced rate due to local queuing strategies and the desired SWARM 1 rate will be propagated by SWARM apportionment to upstream metered ramps Therefore we can judge that the earlier design of SWARM was not implemented If this description is right the queuing strategies should be operated before the calculation of SWARM 1 rate However we cannot find any code in the source code As a result we only implement the local controller strategy in this SWARM plugin 3 Kalman filtering 19 PATH ATMS Center SWARM Ramp metering control Plugin The default forecast lead time is 15 minutes or 30 intervals ahead The number of points used to estimate density slope as input to the Kalman filter is 6 The accumulated density is used in Kalman filtering which has a typical trend like the following 29 60 87 119 150 However the variation in the accuracy of the density measurements is 0 04 allowed range is from 0 02 to 0 5 the variance representing the inaccuracy of the model is 1 0 allowed range is from 0 5 to 10 We think these two default values need to be calibrated 4 2 Contact information Any comments and suggestions are welcome Please contact us at the email address Ichu translab its uci edu 4 3 References 1 Advanced Transportation Management System Traffic Engineer s Manual Revision 1 Prepared by NET fo
12. eeway The corresponding metering rate of the on ramp will be the maximum rate defined in the swarm control 3 6 2 Log sat density Log sat density used to store the calculated saturation density at each mainline detector station The following is an example of the resulting Log sat density file time 405n5 74ml 405n5 55ml 405n4 03ml 405n3 86ml 405n2 99ml 405n1 93ml 405n1 11ml 405n0 93ml 405n0 6ml 06 30 00 45 45 45 45 45 45 35 45 45 time vds name old sat new sat smoothed sat 08 13 30 405n0 93ml 45 45 45 08 15 00 405n2 99ml 45 44 45 08 16 30 405n4 03ml 45 52 45 08 17 00 405n3 86ml 45 45 45 08 19 30 405n1 93ml 45 47 45 08 28 00 405nl 11ml 35 34 35 There are two parts in Log sat density In the first part the first row listed names of all mainline loop detector stations The second row shows the saturation density values of these stations obtained from vds control The second part shows the calculated saturation densities of mainline detector stations The Sample size required to compute saturation density is a user specified variable ranging from 6 to 1200 Its default value is 200 The second part has 5 columns i e the time of calculation name of detector station original saturation density calculated saturation density and the smoothed saturation density The calculation of the smoothed saturation density uses a parameter i e 16 PATH ATMS Center SWARM Ramp metering control Plugin Smoothing factor
13. for saturation density computation defined in the second part of swarm global The default value of this parameter is 0 05 smoothed sat old _ sat 1 0 new sat a If you do not know the saturation density at a detector station we strongly recommend users to use the first simulation run to calculate saturation density values at all detector stations The value shown in the column of new sat is a good value of the saturation density because it does not involve the use of the smoothing factor Then you can update the saturation density values in vds control For a following simulation run the saturation density values shown in vds control are applied before enough data are collected for the calculation of new saturation density values If a new saturation density is calculated the smoothed saturation density will be applied instead of the value shown in vds control As a result the first simulation run will help users know what the saturation density is at mainline detector stations In order to update an inaccurate saturation density users need to edit the vds control file manually 3 7 Error checking Under network directory a file named Log swarm txt is generated for storing all temporary calculations and outputs of SWARM control This file employs two purposes 1 It can be used to check if there is any problem in the global control vds control and swarm control files 2 It can be used to c
14. he following requirements 1 For ramp metering plugin on ramp signals controlled by the SWARM algorithm should be specified in ramp control For example if the on ramp signal 33 is under SWARM control on ramp signal 33 also needs to be specified in ramp control as the following format on ramp signal 33 name 405N amp ICD 1 0 93 14 PATH ATMS Center SWARM Ramp metering control Plugin demand detector 405n0 93orb number of control plans 2 from 6 0 to 9 0 METER_ON with 1 veh per 6 sec from 15 0 to 19 0 METER ON with 1 veh per 6 sec If an on ramp signal is specified in swarm control but not in ramp control this on ramp signal will be regarded as a METER OFF meter 2 For the loop data aggregator plugin all detectors used by SWARM should be specified in loop control for the aggregated data collection In addition the metering rate update interval specified in the second row of swarm control must be the same as the report cycle in loop control 3 6 Output files Each simulation run will generate two output files Log meteringRate and Log sat density They can be found in the subdirectory network Log run xxx where network is the name of the current working directory and xxx is a three digit sequence number The contents of these files are dynamically saved Users can check the progress and condition of the SWARM control during the simulation process 3 6 1 Log meteringRate Log
15. heck if SWARM is operated as expected or if the implementation of SWARM in the target network is proper At each time step the detailed metering rate calculations including SWARM 2a SWARM 2b SWARM 1 are recorded Since this file is saved to the network directory the file is generated based on the latest simulation run 3 8 Fine tuning parameters of SWARM There are many parameters in SWARM During the testing process users can modify all parameters A parameter or the combination of a set of parameters may be appropriate for a network but may not be appropriate for another network 17 PATH ATMS Center SWARM Ramp metering control Plugin Some performance measures are needed to fine tune parameters of SWARM Users can use the measurements file to collect statistical data or use the freeway MOE plugin we developed to collect these measures The saturation density is an important parameter of SWARM The goodness of this parameter affects the performance of SWARM Users can start the first simulation run with the SWARM control with a series of assumed saturation values Then you can check the Log sat_density file under Log directory which is used to record the saturation density values of all mainline loop detector stations in the network calculated based on the second degree polynomial method The value in the column of new sat might be a better value of saturation density that can be used to replace your assumed values
16. m depends on accurate loop detector data 2 2 Development framework Figure 2 illustrates our hierarchical framework to develop advanced ramp metering algorithm plugins in PARAMICS The SWARM plugin is built based on two basic plugin modules i e ramp metering controller and loop data aggregator The onramp signals in the simulation network are controlled by the ramp metering plugin through which metering rates can be queried and set by the SWARM plugin The loop data aggregator emulates the real world loop data collection typically with a thirty second polling interval and broadcast the latest loop data to the dynamic memory At each time increment the SWARM plugin accesses the PATH ATMS Center SWARM Ramp metering control Plugin dynamic memory and obtains the required loop data through interface functions provided by the loop data aggregation plugin Then the metering rate for the next control interval is calculated based on the SWARM 1 algorithm The new metering rate is finally sent back to the ramp controller plugin for implementation PARAMICS simulation New rate Loop Data Ramp metering Aggregator Controller Loop data Old metering rate New rate Advanced metering algorithms Figure 2 The hierarchical approach for the development of advanced metering algorithms 2 3 Pseudo codes Based on the algorithm logic of SWARM and the hierarchical development framework for advaced ramp metering algorithms shown in Figure 2 w
17. meteringRate used to record the metering rates of all on ramps during operation with the interval of metering rate update interval defined in swarm control For example RAMP 07 25 30 07 26 00 07 26 30 07 27 00 07 27 30 07 28 00 07 28 30 07 29 00 07 29 30 07 30 00 07 30 30 07 31 00 07 31 30 07 32 00 J o Os Nn 5 ON o2 ON Oo Oo p ee pak 0 0 0 0 0 e e e e e e e i e i i lA Soo ooo OO OQ De i e A i l Ald Soo ooo OO OG Oe AAAASSARANIANN Sooo De ee Ser aS mer 0 0 Qo t D ww cococudGoodcood WW O9 WwW W pp dro e enone es ese QE e imu ps pen PUE po Un tn CA CA CA mM AAA L A HAmRM AMAA cocco CA CA pus C s S rr 15 PATH ATMS Center SWARM Ramp metering control Plugin 07 32 30 10 10 100 100 151 10 10 300 For each omramp there are a metering rate and a queue override flag outputs The unit of the metering rate is vehicle per minute If the value of the metering rate is 1 it means there is no metering at the on ramp or called greenball The queue override flag can be a value of either 1 or 0 1 represents that the metering at the on ramp is under the default queue override strategy control The default queue override strategy can be described as follows If the occupancy value of the queue detector exceeds a threshold i e 50 the maximum metering rate will be applied to the corresponding meter in order to release more vehicles to fr
18. n the storage zone based on density values of upstream and downstream loop stations Calculate volume change of the storage zone Calculate cumulative storage Calculate storage within a zone Calculate critical storage Calculate maximum available capacity Calculate desired available capacity Calculate metering rate for this ramp Minimum rate control Maximum rate control Rate smoothing Startup shutdown strategy Kalman Filtering For each bottleneck ramp Update observations PATH ATMS Center SWARM Ramp metering control Plugin Time update Measurement update Forecast the density ahead of time FORECAST LEAD TIME j j Determine local max as the upper limit of calculation of SWARM_1 rate Based on the metering mode such as SWARM 2A 2B Traffic apportionment If excess volume at a bottleneck Assign desired reduced traffic to related meters Calculate metering rate for related meters Minimum rate control Maximum rate control Rate smoothing Startup shutdown strategy Check SWARM s start up strategy Find the metering rate based on SWARM s metering mode Queue control is required Implement new metering rate to meters through ramp metering plugin PATH ATMS Center SWARM Ramp metering control Plugin 3 Step by step user manual The SWARM plugin has three input files 1 swarm global containing global parameters of SWARM 2 vds control containing information of network detector stations and bottlenecks
19. nsity 1 Tcrit Excess Density Then the volume reduction at each detector is Volume Reduction Local Density Target Density Number of lanes Distance to next Station These volume reduction values are then distributed to upstream ramps within the defined area of influence for each site using pre defined weighting factors at each ramp based on ramp demand queue storage capacity etc The most restrictive volume reduction is then utilized at each ramp location PATH ATMS Center SWARM Ramp metering control Plugin Density I I I I I l l I I I I I I A i l pathiatibn i eal Density I T l I I I I I I I I I I I I I I I I I I 2 m a a aa A R Sa A A OI XA c Figure 1 SWARM forecasting 2 1 2 SWARM 2 SWARM 2a uses a density function to compute local metering rates based on headway theory Theoretically it attempts to maintain headway at the detector station upstream of metered ramp by optimizing density to maintain maximum flow SWARM 2b introduces a concept of storage zone which starts from the mainline upstream VDS to the next downstream mainline VDS The number of vehicles storing within this storage zone will be calculated Then SWARM 2b computes metering rates to maintain demand such that LOS D as along as possible If there are on ramps and off ramps between the two VDSs detectors are required to be placed at on ramps and off ramps for counting traffic volumes This algorith
20. r Caltrans District 7 June 2000 2 Integrated Ramp Meter Arterial Signal Control Project Detailed Design System Wide Adaptive Ramp Metering Prepared by NET for FHWA FOT City of Irvine and Caltrans Distrct 12 August 30 1996 3 System Wide Adaptive Ramp Metering High Level Design Final Draft Prepared by NET for Caltrans and FHWA June 19 1996 20 SWARM Ramp metering control Plugin PATH ATMS Center Al treana SR 405 IRVINE CENTER Dit OC AM O1EGO CR yer tysvans sep LAGUNA O48 1 498 Vn r Ave 9 e t RET WE JtFFREY RO Q C 3 047 4 742 WB IRVINE CENTER DRE Ene AA b EHE ET aN I PERSRE EE c ppc ME e MC Un ee B maost gr 2 niet See See a oF oleic E 8 EM m LLL EEEE ibe ee eee eee Sere tee E zm Su A e A S Ld i I AM y P Ef iy 1 JES f rt Ta Br n 3 z 1 s k ot E amp d Y E i i Lnd Eog E i od Y 5 n ig ag 3 B i t Mb i oq x A 5 S 14 4 S oz d 1 S 44 JMwy r m s s Joc05 dgn Feb 07 2001 14 57 43 APPENDIX 1 An example of the infrastructure layout obtained from Caltrans 21
21. sion 2 65 0 Free speed 55 0 85 0 range 55 0 to 85 0 precision 1 The second part includes parameters used for the calculation of the saturation density for each loop station 2200 Maximum hourly lane volume 1800 2500 range 1800 to 2500 precision 0 200 Sample size required to compute saturation density 6 1200 range 6 to 1200 precision 0 0 05 Smoothing factor for saturation density computation 0 0 1 0 range 0 0 to 1 0 precision 2 The third part includes parameters of Kalman filtering used in the SWARM 1 algorithm 1 Max number of VDSs to search if bottleneck VDS is failed 1 4 range 1 to 4 precision 0 30 Number of previous time intervals used for the forecast 2 60 range 2 to 60 precision 0 6 Number of points used to estimate slope for Kalman filter 2 9 range 2 to 9 precision 0 30 Number of time intervals into the future to forecast 1 60 range 1 to 60 precision 0 0 04 Variation in the accuracy of the density measurements 0 02 0 5 range 0 02 to 0 5 precision 2 1 0 Variance representing the inaccuracy of the model 0 5 10 range 0 5 to 10 precision 1 The fourth part includes parameters of apportionment algorithm used in the SWARM 1 algorithm 1 0 Fraction of excess traffic to propagate within a section 0 0 1 0 range 0 0 to 1 0 precision 1 0 85 Fraction of excess traffic to propagate between sections 0 0 1 0 range 0 0 to 1 0 precision 1 The fifth part includes parameters of the
22. t file named Log sat_density Since the network model may not be calibrated well the saturation density calculated based on real world loop data might not be the same as that calculated based on simulation 3 4 Preparation of swarm_control This file includes the design of the SWARM algorithm to the target network total number of SWARM controlled ramps is 8 metering rate update interval 30 report metering rate yes ramp 92 freeway 405 direction N postmile 5 74 mainline detector 405n5 74ml upstream ramp 95 N A N A N A apportionment factor 1 0 0 0 0 0 0 0 onramp detector 405n5 74orb queue detector 405n5 74orspill HOV 0 metering mode SWARM 1 minimum rate control ABS MIN default rate control ABS MAX swarm startup strategy RUN SWARM DURING TOD ONLY rate restriction 6 30 The unit of the metering rate in swarm control is veh minute The on ramps such as 02 listed in swarm control are ordered by freeway id direction from downstream to upstream The furthest downstream ramp on a freeway is input first mainline detector is the corresponding mainline detector of a ramp postmile can be obtained from Caltrans which is a basic input of SWARM In general a ramp s postmile is the same as the postmile of the corresponding mainline detector of the ramp upstream ramp and apportionment factor are used for metering rate apportionment the second part of the SWARM 1 algorithm metering mod
23. tector station shown in Figure 3 If your simulated network in the real world does not have the required detector configuration for the SWARM control please add demand detector on ramp detector mainline detector and queue detector to the simulation network based on Figure 3 in order to correctly use this plugin 3 2 Preparation of swarm global SWARM has several global parameters They are defined in the swarm global file Totally there are 27 parameters The first part of this file is about loop data acquisition Please see Section 2 6 of reference 1 for more detailed description The value shown on the leftmost side of each line is actually the recommended value 25 0 Ave veh length for the right lane 10 0 30 0 range 10 0 to 30 0 precision 1 22 0 Ave veh length for the 2nd right lane 10 0 30 0 range 10 0 to 30 0 precision 1 18 0 Ave veh length for other ML and HOV lanes 10 0 30 0 range 10 0 to 30 0 precision 1 10 PATH ATMS Center SWARM Ramp metering control Plugin 9 0 Absolute maximum effective loop length 6 0 10 0 range 6 0 to 10 0 precision 1 9 Volume threshold to compute a new effective loop length 4 14 range 4 to 14 precision 0 3 0 Absolute minimum effective loop length 2 0 6 0 range 2 0 to 6 0 precision 1 55 Speed threshold to compute a new effective loop length 50 90 range 50 to 90 precision 0 0 02 Smoothing factor for the effective loop length 0 0 1 0 range 0 0 to 1 0 preci
Download Pdf Manuals
Related Search