After completing Tasks 1-5 of the Miasma Beach Transportation Model, you will focus on improving the overall model system in three ways:

1. Task 6. Model Critique
2. Task 7. Model Feedback
3. Task 8. Model Extension/Enhancement

A final Task 9 will provide a final report, a final presentation, and a set of methodology lecture slides. There will be several intermediate deliverables.

Miasma Beach Tasks 6-9.

Preferred Projects for Spring 2018
The following model extensions/enhancements are recommended for this year. One of these projects will be assigned to you unless an independent project proposal is submitted and accepted.

1. A Macroscopic Analysis of a Transit System for Miasma Beach (TransCAD)
2. A Disaggregate Trip Generation Data Set and Category Model for Miasma Beach
3. Mode Choice Models for Miasma Beach: With and Without Transit
4. Air Quality Modeling for Miasma Beach with EMFAC
5. Dynamic Traffic Asignment for Miasma Beach

Overview
The basic Four Step Model, as presented in CEE 123 Transportation Systems III: Planning and Forecasting and as utilized in practice since the 1960s, can be improved in a variety ways, while maintaining the basic structure and approach. Potential areas for improvement include:

1. Network and Activity Data Structure
   Pre/Post Processing [Land Use Models, AQ Analysis, etc.]
   Alternatives Analysis and Economic Evaluation
2. Trip Generation
3. Trip Distribution
4. Mode Choice, Transit Analysis, and/or Matrix Adjustments
5. Trip Assignment
6. Microsimulation
7. Tour- and Activity-based Models
8. Software Platforms
9. Freight Transportation

Basic Analysis
The CEE 123 Miasma Beach Project will be used as a baseline reference model and data set. This model system generates daily person trips and produces estimates of mode- and period-specific volumes and travel times on the Miasma Beach transportation network. TransCAD has been used as the base software platform, but other software applications are available). After the base Miasma Beach analysis is completed (Tasks 1-5), a model enhancement will be developed, integrated with the base model, and fully documented. Both the base model applicaiton, and the enhancement, must include model feedback.

Model Enhancements and Extensions
The following summary presents potential model enhancements and extensions organized by overall process improvements or, where applicable, by process step. Several of these improvements are relatively straight-forward, and may need to be completed as a set of model improvements. On the other hand, a few of these improvements are substantial and may need to be completed in a small group. Each of these improvements should be completed jointly with feedback in the 4-step model process.

Many of these projects have been addressed in prior years, to varying degrees of completion. If you select one of these topics, you may choose to build upon the work already completed.

1. Overall Process
   The process is designed to produce an equilibrium solution, but in virtually all application prior to the last 5-10 years, the only equilibration occurred in the Route Choice stage where, effectively, static demand matrices were assigned to the network. While direct feedback of resulting link travel times to prior stages is fairly easy to implement, this process will not converge. There are several feedback methods available.
   1. Improvement 1A
      Introduce feedback into the basic model by feeding link travel times from Route Choice back to skim tree generation for Trip Distribution. Consider Evans' algorithm or MSA; review relevant literature.
      
      

      All projects must introduce at least a minimum level of feedback, in lieu of the transportation system design component of CEE 123 (Task 6).

      • Alternate Methods of Feedback to Trip Distribution (Hao Yang, Spring 2009)

   2. Improvement 1B
      Introduce feedback to Trip Generation by developing accessibility measures for each TAZ/HH location which would influence productions (and/or attractions), thus influencing the total amount of travel in the region as a function of congestion/accessibility.
      • Feedback in the FSM incorporating Accessibility in Trip Generation (Pedro Camargo, 2010)

   3. Improvement 1C
      Introduce an Activity System Forecasting component fronting the four-step model (producing the activity system variables required for input to the Trip Generation models). Ideally, the feedback of travel times from Route Choice will extend to the Activity System model. Land Use (Activity System) models are an explicit phase of the Travel Model Improvement Program (TMIP) and several models are available for your use.
      • A Land Use Forecasting For Miasma Beach (P.McGowen, Spring 2003)
      • Linking a Garin-Lowry Land Use Model with a TransCAD-based Four Step Model (M.Allahviranloo, 2010)

   4. Improvement 1D
      Develop a formal evaluation component to enable the comparative assessment of the performance of system network/activity alternatives using Benefit/Cost or other evaluation methodologies.
      • A Benefit/Cost Methodology for Alternatives Analysis (Raquel Girvin, Spring 2003)
      • Benefit/Cost Analysis (Jeff Wang, Spring 2006)
      • Benefit/Cost Analysis (Neil Smolen, Winter 2014)

   5. Improvement 1E
      Add an air quality modeling and/or other post-processing component to the model system.
      • Evaluation of Vehicle Emissions with Paramics Microsimulation (YounShik Chung, Spring 2003)
      • Linking MOBILE5 to the Miasma Beach Transportation Model (K.Nesamani, Spring 05)
      • Linking MOBILE6 to the Miasma Beach Transportation Model (C.Long, Spring 09)
      • Evaluation of Vehicle Emissions with TransModeler Microsimulation in Miasma Beach Network (Qijian Gan, 2010)
      • Air Quality Analysis using MOVES (Y.Bae)
      • Air Quality Analysis using EMFAC2017 (M.Ramirez Ibarra, Spr'18)
      • Vehicle Class Data and Multi-class Assignment (B.Casebolt, Spr'18)

   6. Improvement 1F
      Add an evaluation of intersection performance, by utilizing TransCAD ICU analysis or by linking TransCAD with Syncho or other package. Consider adding turn penalties and prohibitions to selected intersections or use an advanced assignment procedure in TransCAD.
      • Linking to Synchro for Intersection Post-processing (Miriam Leung, 2003)
      • Integrating Synchro Delay Estimates Feedback to TransCAD (Albert Chung, Spring 2006)
      • Using ICU Analysis in TransCAD (Lawrence Liu, Spring 2006)
      • Using Synchro Delay Estimates to Compute Turn Penalties for TransCAD (Ali Yavari, 2009)
      • Linking Synchro to TransCAD and Increasing Network Density on Coast Highway (Paul Harer and Andrew Polgar, 2010)
      • Linking Synchro to TransCAD (Yiqiao Li)
      • Using Synchro to Determine Turn Penalities in TransCAD (Steven Ng, S'18)

   7. Improvement 1G
      Increase the density of network links to reflect local streets, more analysis zones, or other supply enhancements. Network enhancements should reflect both secondary and collector/local street within the base primary/secondary grid. Zoning system enhancements should focus on increasing the number of zones. A relatively simple change would be to split TAZ 3 and/or 4, adding appropriate network links and centroid connectors. Other changes include adding network links south of Coast Highway and reflecting TAZ access to these links. An alternate project would be to increase the density of network links to reflect local and collector streets or other network enhancements.
      • Increasing the Number of Analysis Zones (Sarah Aly, Spring 2006)
      • Network Improvement and Evaluation (J Paul Townley, Spring 2007)
      • Linking Synchro to TransCAD and Increasing Network Density on Coast Highway (Paul Harer & Andrew Polgar, 2010)

   8. Improvement 1H
      Sketch Planning Models:
      • Apply the 3Ds in Miasma Beach (C.Kwan and Shangyou Zeng, Winter 2014)

2. Trip Generation
   The standard regional model application utilizes category models for productions and regression models for attractions (and applies standard trip balancing by purpose).
   1. Improvement 2A
      Replace production regression models with Category Models using a simulated disaggregate data set consistent with Miasma Beach aggregate data. This will require generating a household data set consistent with aggregate Miasma Beach TAZ data set. Software to generate this data is available.
      • A Disaggregate Data Set and a HH Productions Category Model (Junping Duan and Liping Gan, Spr'03)
      • Synthesizing Population Data for Miasma Beach (R.Rafiq and K.Allu)
      • Synthesizing Population Data for Trip Generation and Mode Choice (F.Khatun and T.Ahmed, S'18)

   2. Improvement 2B
      Replace production category models with Multiple Classification Analysis (MCA) models (see McDonald and Stopher, 1983). This will require generating a household data set consistent with aggregate Miasma Beach TAZ data set. Software to generate this data is available.
      • Multiple Classification Analysis for Miasma Beach (Lesley Wang, Spr'04)
      • Synthesizing Population Data for Trip Generation and Mode Choice (F.Khatun and T.Ahmed, S'18)

   3. Improvement 2C
      Replace production category models with Person Category Models (see Supernak et al., 1983). This will require generating a person-level data set consistent with aggregate Miasma Beach TAZ data.
      • A Person-category Trip Generation Model (Xing Zheng, Spring 2006)

   4. Improvement 2D
      Develop a methodology for forecasting joint distributions of future activity system variables required for category-type generation models. Alternatives include Interative Proportional Fitting (see Beckman et al., 1996), conventional multiple equation approaches (OCTAM and SCAG), and general linear modeling (GLIM). Essentially, these methods will produce a joint distribution of households given marginal distributions of activity characteristics.

   5. Improvement 2E
      Develop discrete choice based trip generation models.
      • A Discrete Choice Trip Generation Model

   6. Improvement 2F
      Consider land use trip rate approaches integrated with a GIS in place of category or regression model approaches. You will need to generate land use data for Miasma Beach that, together with real or reasonable trip rates, are consistent with aggregate trip generation in Miasma Beach.
      • A Land-Use Based Trip Generation Model

   7. Improvement 2G
      Develop reallocation models to reassign generated productions for non-home-based (NHB) trips from the zone of residence to the zone of production/origin.

3. Trip Distribution
   The standard regional model application utilizes singly constrained gravity models for internal trip distribution. These models often use friction factors that are subsequently smoothed instead of direct estimation of these continuous functions. Also, the use of doubly constrained models has been relatively rare despite an explicit need for both production and attraction constraints. Instead, ad hoc methods for balancing the resultant matrices (via Attraction Balancing or Column & Row Factoring (Furness)) are introduced.
   1. Improvement 3A
      Develop required data to calibrate the Miasma Beach trip distribution model (for at least the three trip purposes). There is some variance allowed for the resulting trip tables, as long as the entire Miasma Beach model system is consistent (this allowance is also valid for the other trip distribution improvements).
      • Generating Travel Survey Data and Calibrating Trip Distribution Models (HKKim, Spring 2003)

   2. Improvement 3B
      Introduce and compare singly constrained gravity models in place of doubly constrained models and utilize direct estimation of "smooth" impedance functions (e.g., power, negative exponential, and combined) versus friction factors.

   3. Improvement 3C
      Introduce and compare multinomial logit models of destination choice in place of doubly constrained gravity models (easily introduced in TransCAD).
      • Developing a Discrete Destination Choice Model (Kyra Tao, Spring 2004)

   4. Improvement 3D
      Develop a (growth factor) forecasting model for external trips in Miasma Beach. Introduce an approach to simultaneously balance internal and external trip generation as input to distribution models.

   5. Improvement 3E
      Develop a intervening opportunity distribution model in place of the doubly constrained gravity models (easily introduced in TransCAD).

4. Mode Choice
   The current version of the Miasma Beach exercise does not incorporate a mode choice component. Rather, person trips are converted to vehicle trips using average vehicle occupancy which reflects the ratio of total person trips to total vehicle trips.
   1. Improvement 4A
      Develop a (or enhance the current elementary) transit network for Miasma Beach. There can be limited overall service but at least two separate but coordinated routes.
      • Developing a Transit Network for Miasma Beach (Jamie Lai, Spring 2001)
      • Transit Network Assignment (Palm Apivatanagul, Spring 2004)
      • Transit Modeling in TransCAD (Neelam Sharma, Spring 2009)
      • Developing a Transit Network in Miasma Beach (Sarah Tasmin & Shayesteh Vafai, Winter 2014)
      • Transit Network Development for Miasma Beach (N.Sharmeen)
      • Transit Network Development and Modeling for Miasma Beach (J.Kwong, Spr'18)

   2. Improvement 4B
      Introduce a multinomial (or hierarchical) logit model for mode choice. This may require development of a transit network for the Miasma Beach exercise (or work jointly with others). An alternative would be to model drive alone versus carpool and introduce time and cost differences.
      • Developing a Diaggregate Data Set and Discrete Mode Choice Model for Miasma Beach (Veronica Alvarez, Spring 2001)
      • A Diaggregate Data Set and Discrete Mode Choice Model for Miasma Beach (Roberto Ayalya, Spring 2009)
      • Promoting Public Transit: A Nested Logit Model Application (Jielin Sun, 2010)
      • A Mode Choice Model for Miasma Beach (P.Ahmed, M-T Ho, and J.Yu, Winter 2014)
      • Mode Choice Models for Miasma Beach (J Tao)

   3. Improvement 4C
      Generate a disaggregate mode choice data set (preferably consistent with overall Miasma Beach trip generation).
      • Developing a Diaggregate Data Set and Discrete Mode Choice Model for Miasma Beach (Veronica Alvarez, Spring 2001)
      • Synthesizing Population Data for Trip Generation and Mode Choice (F.Khatun and T.Ahmed, Spr'18)

   4. Improvement 4D
      Develop a complex transit system network and also introduce a simple mode choice model to generate modal trip matrices for assignment; or develop a complex mode choice model (any set of modes), perhaps building a nested model, but skip transit assignment (e.g., use a drive alone versus carpool mode choice model and use this to determine vehicle trip matrix for highway assignment; or both do complex tasks jointly with another student.

5. Trip Assignment
   The current Miasma Beach exercise utilizes only standard UE assignment via the Frank-Wolfe algorithm. Only highway trips are assigned.
   1. Improvement 5A
      Introduce (a) stochastic assignment, (b) stochastic UE, or (c) other assignment methodologies in place of standard UE. This alternative is not available to those who have completed a similar exercise for CEE 228A.
      • Alternate Traffic Assignment Models (Arwa Aweiss, Spring 2003)
      • Stochastic User Equilibrium (Oluseyi Ojuri, Winter 2014)
      • Path-based and Origin-based Trip Asignment (Xinyun Cao & Yang Han, Winter 2014)

   2. Improvement 5B
      Introduce alternate link performance functions or alternate parameters for the standard BPR performance functions. Generate a data set on which to base the estimation of these functions.
      • Estimating and Applying Alternate Link Performance Functions (Jiyoung Park, Spring 2003)

   3. Improvement 5C
      Develop and apply multi-class assignment models.
      • Multimodal Multiclass Assignment (Cheng Gong & Guoyi Ruan, Winter 2014)
      • Vehicle Class Data and Multi-class Assignment (B.Casebolt, Spr'18)

   4. Improvement 5D
      Develop and apply Dynamic Traffic Assignment models for Miasma Beach.
      • Dynamic Traffic Assignment for Miasma Beach (Zhe Sun, Spring 2011)
      • Dynamic Traffic Assignment for Miasma Beach (Siming Wang & Qinglong Yan, Winter 2014)
      • Dynamic Traffic Assignment for Miasma Beach (C Qin & X Wang, Spring 2015)
      • Dynamic Traffic Assignment for Miasma Beach (SH Peng & Z Yu, Spring 2015)
      • Dynamic Traffic Assignment for Miasma Beach (Sunny An, Spring 2016)

6. A Microsimulation Model
   The conventional forecasting model, sometimes referred to as a macro-simulation model, is used to produce demand estimates for traffic microsimulation models (such as Paramics) which allow for a more detailed network assessment.
   1. Improvement 6A
      Develop a Paramics (or other) microsimulation model for Miasma Beach. Generate required geometric and control data for the network.
      • A Paramics Network for Miasma Beach (A.Tok and Y.Zhang, Spring 2003)
      • Using Paramics and TransCAD to Model City Bypass Options (Tyler Bonstead, Spring 2005)
      • Miasma Beach Project with TransCAD and Paramics (Gunwoo Lee, Spring 2007)

   2. Improvement 6B
      Develop a TransModeler microsimulation model linked with TransCAD (or a Dynasim microsimulation model linked with CUBE) for Miasma Beach. Generate required geometric and control data for the network. Update the static trip tables from TransCAD with generated link traffic counts.
      • Linking TransCAD and TransModeler in Miasma Beach (Yosuke Arai, Spring 2009)

   3. Improvement 6C
      Develop a microscopic simulation model (TransModeler) for a Miasma Beach public transit system operation. Update the static trip tables from TransCAD.
      • Transit Microsimulation for Miasma Beach (Seth Contreras, Spring 2011)

   4. Improvement 6D
      Develop a mesoscopic simulation model (Dynasmart) for Miasma Beach. Generate geometric and control data for the network. Update the static trip tables from TransCAD.

7. Develop a Tour-based Model Alternative to the Conventional Miasma Beach Model
   Replace the Generation/Distribution/Mode/Time-of-Day components from the Miasma Beach model to a tour-based formulation.
   • Testing MATSim for Miasma Beach (M.Mosslemi, Spr'18)

8. Convert Miasma Beach Model Software Platform
   Port a version of the Miasma Beach model from TransCAD to an alternate modeling platform. The Miasma Beach Tranplan version has not been updated in several years and we now have a license for Citilabs CUBE. You will create an equivalent and fully-function version of Miasma Beach, both data sets and documentation, for the alternate software platform.
   1. Improvement 7A
      Conversion to Citilabs' CUBE Platform:
      • Miasma Beach: Conversion from TransCAD to the Cube Platform (Joe Chow, Spring 2007)

   2. Improvement 7B
      Conversion to TRANSIMS Platform:
      • Miasma Beach: Conversion from TransCAD to TRANSIMS (Jamie Kang, Spring 2007)
      • Miasma Beach: Conversion from TransCAD to TRANSIMS (Miyuan Zhao, Spring 2010)

   3. Improvement 7C
      
      • Conversion to INRO's EMME/2 Platform:

9. Develop a Freight Transportation Model for Miasma Beach
   Develop data sets and revised networks to add freight components to the Miasma Beach model, such as a small port, freight forwarding facilities, external freight traffic, etc.
   1. Improvement 8A
      
      • A Freight Transportation Model for Miasma Beach (Fatemeh Ranaiefar, 2010)
      • Freight Transportation Modeling (Junhyeong Park & Yue Sun, Winter 2014)

Miasma Beach Project
Variations of the Miasma Beach exercise have been utilized as the term project for the last several years. The preferred data set is the Winter 2017 TransCAD version of this exercise, unless otherwise approved.