Resume

John A. Gibb

Senior Transportation Engineer

DKS Associates, Sacramento, since 1990

Areas of Expertise

  • Travel Model Development, Calibration, Troubleshooting, Specialized Applications
  • Travel Model Application
  • Travel Survey Analysis and Model Estimation

California Registered Professional Traffic Engineer No. 1627

Member, Institute of Transportation Engineers

Bachelor of Science, Civil Engineering, California State University, Sacramento, 1987

Publications

  • Solving Travel Demand Model Equilibrium with Barzilai-Borwein Step Sizes, presented at TRB 2016 Annual Meeting
  • “Adaptive Stopping Criteria for Traffic Assignments within Convergent Feedback Demand Models,” TRB 2015 Annual Meeting
  • “Model of Traffic Flow Constraint through Nodes for Dynamic Network Loading with Queue Spillback,” 2011, Transportation Research Record No. 2263
  • “Convergence of an Activity-Based Travel Model System to Equilibrium: Experimental Designs and Findings,” Transportation Research Board (TRB) Planning Applications Conference Daytona Beach, Florida, May 2007
  • "Disaggregate Quasi-Dynamic Park-And-Ride Lot Choice Models With Parking Capacities," TRB Planning Applications Conference, Houston, May 2009
  • "Network Assignment and Equilibrium for Disaggregate Models," TRB Planning Applications Conference, Houston, May 2009

John Gibb specializes in the development and application of travel demand models for regional planning, traffic, transit, and multimodal engineering and planning. He has developed, calibrated and applied travel demand models of numerous cities and regions throughout the U.S., and has used specialized statistical techniques to describe, estimate and calibrate models of travel behavior from travel surveys. He is experienced with the Cube Voyager, TransCad, Visum, EMME (and Minutp, TranPlan) software packages for model development and application, and ALogit, SPSS, and Limdep for statistical description, hypothesis testing, and model estimation, as well as Access and other database programs for efficient and complex data manipulations, and Basic and Visual Basic programming (and even Fortran).

Travel Model Development

Developed the application framework for the Sacramento regional activity-based travel demand model using DaySim (Bowman and Bradley), with Cube-Voyager scripts integrating trip matrix processing, traffic and transit network assignment and level-of-service “skim” calculations with tolling, capacity-constrained park-and-ride lot choice, auxiliary trip models including airport passenger ground access, and efficient, innovative, verified equilibration feedback techniques.

Previously, updated and re-validated the SACOG four-step model based on year 2000 household survey data for the Sacramento region. Estimated and calibrated the original 1995 version of that model from 1990 household survey data.

For Kern County COG, made improvements, enhancements, updates, and base-year recalibration to Kern County model’s gateway data, trip generation, trip distribution and mode choice models, and post-model analysis. Compiled gateway through and internal-external traffic flows from newest California Statewide Travel Demand Model, and applied them as model inputs, adjusting to traffic counts. Diagnosed and corrected computational problems in trip distribution, and adjusted trip generation and mode choice parameters and inputs toward agreement with California Statewide Travel Survey, transit boarding counts, and traffic counts.

Upgraded the Spokane Regional Transportation Council’s Visum travel demand model, fitting trip generation, distribution, mode choice, and time-of-day components to a regional household travel survey, and calibrating to meet reasonableness comparisons to transit boardings and traffic counts.

Developed the most recent application framework for the Shasta County Activity-Based Model, integrating DaySim with Cube networks, based on the Sacramento application. Developed external trip data from California Statewide Travel Demand Model.

Performed several updates, expansions, and enhancements to the Carson City MPO model, currently in TransCAD (and longer ago a TranPlan version). Most recently, added mode choice to the model, and equilibrium feedback. Developed external trip data from AirSage queries.

Longer ago, developed MINUTP travel demand models for both Fresno County and Shasta County using household travel survey data and roadside survey data. Also provided linkages between these MINUTP models and the DTIM air quality analysis program.

Project engineer for the development of travel models for Alameda, Placer, Fresno and Shasta counties, and for the cities of Vacaville, Roseville, Rocklin and Redding. For these efforts, developed trip generation system from detailed land use data, calibrated models to current conditions, prepared forecasts of future scenarios, and provided training to agency staff on the models.

Travel Survey Analysis and Model Estimation

For Oregon Dept. of Transportation, estimated components to a commercial vehicle demand model. Processed, filtered, and interpreted trip diary records from a survey in Ohio to estimate trip generation, distribution, mode choice, and time-of-day parameters, tentatively assuming transferability.

Estimated the component models of the previous trip-based models of the Sacramento and Puget Sound regions. Using data from household travel diary surveys, developed logit models of mode choice, workplace and destination choice (trip distribution), including a combined nested workplace-mode choice model. Estimated and evaluated cross-classification models of household trip. Studied relationships between demographics and trip generation, using analysis of variance (ANOVA) and similar techniques. Also analyzed trip attraction using least-squares and maximum-likelihood methods.

Travel Model Application

Applied regional model for two LRT extension projects following FTA guidelines. For the South Sacramento LRT Corridor AA/EIS, revalidated SACOG’s regional model and received FTA approval of the model and forecasting methodology. Applied the model for forecasting future transit ridership under baseline, TSM and LRT alternatives and provided FTA criteria for new-starts. Calculated user-benefits for various project alternatives using FTA Summit program. Developed extraction routines to provide the needed data from the mode choice model. Also applied regional model for the Folsom Corridor LRT EIS project.

Applied regional models for two multi-modal MISs. For the I-80 Corridor MIS, applied model to evaluate new HOV lanes, LRT and BART extensions, express bus service on HOV lanes, commuter rail service and new ferry service in a 65-mile long corridor. Also applied regional model for the U.S. 50 Corridor MIS to test phased implementation of HOV lanes and LRT extensions in a 27-mile long corridor.

Adapted SACOG’s regional model to high detail for analysis of alternatives for the proposed Placer Parkway. Assembled data from SACOG and Placer County’s model. Applied this model to evaluate transit system alternatives in Placer County.

Applied models to develop and evaluate general plans, master plans and circulation plans for Sacramento County, Placer County, Fresno County, the cities of Roseville, Rocklin, Redding and Carson City.

Evaluated and recommended improvements to the City of Seattle’s EMME model concerning its performance in trip generation, distribution, and mode choice, for their Alaska Way Viaduct Replacement Study.

Specialized Modeling Applications

Added a multinomial logit park-and-ride lot choice model with capacity constraints into the Alameda County travel demand model’s auto-access transit demand and assignment components, to support studies of alternative satellite lots and shuttle services.

For the City of Sacramento, developed traffic assignment with capacity-constrained parking location choice with city-block detail covering the greater downtown.

For a Caltrans funded research effort, developed a model of the choice for carpool drivers to use or not to use an available carpool lane on a freeway. For this effort, estimated logit models, and used general statistical and database techniques, upon carpool trip-survey and traffic observation data. Applied this model within the Sacramento area regional model system's equilibrium traffic assignment for a multi-modal corridor study that evaluated light rail extensions and HOV lanes. Also developed a new linkage between SACOG's regional model and DTIM that reduces model run time and storage requirements.

Synthesized multi-dimensional household cross-classification from US Census PUMS and marginal tables using iterative proportional factoring method for various models including Kern County CA, Bellevue WA, Carson City NV, and Spokane WA.

Experienced with modeling techniques and issues associated with trip generation from parcel data, select-link-based cost-sharing analysis, peak hour and other time-of-day modeling; study area focusing and subarea extraction, impact of bottlenecks upon traffic flows, and intersection and ramp-meter delays in traffic assignment.

Independent Ongoing Research.

Identified and tested step size formulas for solving for equilibrium of a model system with feedback more efficiently than “method of successive averages” and without trial-and-error as with constant step sizes. The Barzilai-Borwein step sizes are computed very quickly during runtime, immediately before each iteration’s use, from the linear trend of the current and previous iterates. Developing an adaptation of this method for random noise in the iterates, with early results presented at the 2016 Innovations in Travel Modeling conference.

Measured the error of skim matrices due to the level of the preceding assignment’s convergence (in relation to nearly perfectly-converged assignments), with varying software, assignment algorithms, and regions. From this information, identified a means to choose assignment stopping criteria so that assignments may be sufficiently converged to support feedback convergence, but not more than necessary. This strategy reduces run-times – particularly of assignments within the early feedback iterations when skim times fluctuate anyway.

In another research initiative, created a macroscopic capacity-constraint model for generic intersections with spillback. Dynamic traffic assignment models require such a model if they are to propagate queue spillbacks in a network. All previously-published macroscopic node models had problems in causality and consistency. This model represents a first-come first-served behavior among vehicles at the face of the intersection; it was shown to agree reasonably well with microscopic simulation of this behavior, and remains the only node model shown to agree with any simulated behavior.