TRANSPORTATION MODELLING IN CALGARY Why Do We Use Transportation Models? There are three approaches that can be used for proceeding to develop a transportation system for Calgary s future. One approach is to just try something out to see if it works. This is like running an enormous trial and error experiment with the entire city. The public would have to bear the consequences of the errors in the trial and error process and the cost of the errors could be astronomical. The second approach is to use mental models. The user of a mental model develops their own conceptions of the system and forecasts the results of possible actions in their own head. Mental models can sometimes be very good, but they have limitations. They are subject to all the human limitations: they can be slow, easily distracted, distorted, and inaccurate. Such limitations can be severe handicaps when considering a system as complex as the entire city. Also, mental models are difficult to communicate or discuss objectively. There can be as many different mental models as there are people involved. The limitations of the trial and error process and the mental model process have lead to the use of mathematical computer models for mimicking transportation systems and the behaviour of people within them. These mathematical models have many benefits that make their use attractive. They are formal, and objective, and bring out into the open for discussion, all of the assumptions and compromises being made. Mathematical computer models are, nevertheless, only a reproduction of the actual transportation system. They are merely meant to behave in the same way as the real system. They do not tell us what to do they do not give us any instruction. The computer models we use in transportation engineering and planning do not think. They are designed to allow us to ask What will happen if we do this? to allow us to do our trial and error experimentation on a computer rather than in real life. 1 Transportation computer models are an indispensable tool for transportation planning. Overview of Current Models The City of Calgary has operated a computer-based transportation forecasting model since 1964. This travel demand model has been rebuilt and updated several times over the intervening years. It is a model that is used for strategic planning, infrastructure planning, and as input for more detailed analysis. In 1990, The City introduced a second kind of model sub area traffic assignment models. The purpose of these models was to provide more detailed analysis than was 1 Paraphrased from J. D. Hunt, The Role of Modelling in Transportation Planning, for the City of Calgary as background documentation for GoPlan.
available from the city-wide model. These models were applied to community studies, site impact studies and geometric design studies. A couple of years ago, The City experimented with a third kind of model micro simulation (dynamic traffic assignment) models. These models can be applied in much the same way as a traditional traffic assignment model. Since they operate at a very fine level of detail, they can be used for some very complex traffic operation studies. They typically also provide a feature to allow the output to be displayed graphically (animation). In 2001, a couple of large projects began using micro simulation models. Further details regarding the three types of models being applied in The City of Calgary are given below. Regional Transportation Model (RTM) The City of Calgary s travel demand forecasting model is currently named The Calgary Regional Transportation Model and is referred to as the RTM. The software used for this model has changed a few times over the years and is currently the emme/2 software. The current RTM covers the geographic area of the city and some additional lands around the city limits. It is an a.m. peak hour model that represents the major trip purposes and modes of travel occurring in that time period. It deals primarily with personal travel and does not include a representation of trucks. The RTM is a travel demand model. The base information in the model includes land uses (population and employment), travel behaviour and demographic data, and a representation of the transportation network (road and transit). The mathematical structure of the model is such that it can be used to forecast through a 4-stage process where the model: Generates the demand for travel; Distributes the trips from zone to zone; Splits the trips according to mode of travel; and Assigns the trips to the road and transit networks. The RTM is the basic and essential tool for transportation forecasting. The information produced by the model is sometimes used fairly directly and sometimes undergoes further analysis. Example projects where the results of the RTM were a key piece of analysis include: The Calgary Transportation Plan (GoPlan); Transportation Infrastructure Investment Plan (TIIP); GRAMPS; Inner City Transportation Study; 14 th Street S.W., Deerfoot Trail, Stoney Trail, and Macleod Trail South Functional Studies; small area development impact studies; Centre Street Bridge closure;
CFB and Bow Valley Centre; traffic noise studies; and as input to more detailed levels of modelling and analysis. The RTM is currently being expanded and updated. An update is done on a regular cycle (1971, 1981, 1991, and 2001) to ensure that the model remains relevant and reflects current travel behaviour. The model is also being expanded to ensure that it can respond to the increasingly complex and diverse issues facing transportation planning. More information on the RTM Update is included in the section titled Changes in Progress. Please also see the article entitled New Directions for the Calgary Regional Transportation Model for more information on the Update. Subarea Models The purpose of subarea models is to provide analysis at a more detailed level than is practical in the RTM (a model that covers a geographic area as large as the city of Calgary). Subarea models are often, although not always, assignment models. Generation, distribution and mode split are fixed inputs for these models. They do only the last step of the four stages listed above for the RTM. In order to produce forecasts that result from regional changes in land use or transportation systems, the subarea models need information from the RTM. Sub Area (Using a Continuous Traffic Assignment Model) In about 1990, The City of Calgary implemented a second type of modelling in order to produce more detailed analysis than was possible using the RTM. The software chosen was CONTRAM (CONTinous Traffic Assignment Model). In this type of model, the timing and operation of signals can be explicitly represented and so these models are capable of a good representation of the operating conditions in the peak hours at a fairly good level of detail. Typically, these subarea models were set up for the a.m. and p.m. peak hours and covered from 6 to 18 square miles of area. In order to respond to changes in the rest of the city, these models use input from the RTM at the boundaries of the sub area. Subarea models using CONTRAM have been used for a variety of community, land development, capital programming, geometric design and operational studies over the years. A few example projects include: Calgary North ASP; Sunridge development proposals; CFB West; Bow Valley Centre; Centre Street Bridge Closure; Inner City Transportation Study; Transportation Improvement Priority Study (TIPS forerunner of TIIP); North Hill Shopping Centre redevelopment proposal;
Cominco lands development proposal; and Anderson Road / Macleod Trail interchange functional design study. There are currently 14 subarea models using CONTRAM and covering most of the city. However, due to limited resources and higher priority work over the past several years, the models have not been kept updated and many of them are now five or more years old. Sub Area (Using Part of a Demand Model) In 2001, the emme/2 software was used to establish two subarea models. One was for the area generally south of Fish Creek and 130 Avenue S and one was for the NE area of the city. The purpose of these models is to provide more detail than is available from the RTM. However, since the models will be used primarily for long range planning, the operational level of detail available in a continuous traffic assignment model or a micro simulation model is not required. These models were established for the a.m. and p.m. peak periods and will serve as input to manual analysis and/or more detailed modelling analysis and policy decisions. In order to produce these models, the network is coded in finer detail than in the RTM (although these models are not capable of explicitly representing the operation of signals). Information from the RTM is input to advise the subarea models of changes in the city at large. A similar type of model to the ones described above will be set up for Northwest Calgary generally north of Stoney Trail in 2002. Another type of 4-stage subarea model was set up for the built-up area of the Northwest in 2001. The Northwest/University model used the TModel software and was used to address land use and transportation planning issues associated with the development of the University lands and other institutional and commercial sites in the area. Sub Area (Using a Micro Simulation Model) Prior to 2001, the City of Calgary had limited experience with micro simulation models. Using the VISSIM software, the Forecasting Division had done only two small projects: The operation of Edmonton Trail and Memorial Drive during the Centre Street Bridge closure; and The operation of 16 Avenue N.W. adjacent to the North Hill Shopping Centre to investigate the operation of a proposed traffic signal. Neither of the above projects used the dynamic traffic assignment capabilities of the software as this feature has only been available in the most recent releases of the software.
In 2001, the Forecasting Division began work on three projects using micro simulation (Vissim) software: Memorial Drive NW. This is a small network covering the area along Memorial Drive and a few blocks north from Crowchild Trail to 10 Street NW. The model will be used to investigate alternative interchange designs for the intersection of Memorial Drive / 14 Street and to look at transit operations along 10 Street NW in the Louise Crossing area. Shawnessy. This is a large micro simulation network extending from Fish Creek south to Highway 22X and including the commercial areas east and west of Macleod Trail. The model will be used to determine whether any operational improvements are possible to alleviate the congestion around the commercial centres. 11/12 Avenue S. This is a large network covering 10 to 13 Avenues S from Crowchild Trail to the Elbow River. The network will be used to investigate the benefits and costs of converting the 11/12 Avenue one-way couplet to two-way operation on both avenues. Micro simulation models operate at a level of detail even greater than continuous traffic assignment models (CONTRAM and others). They model the car following and lane change behaviour of drivers on a second-by-second basis. Typically, micro simulation models provide visual output that show the individual vehicles moving through the network in animation. Micro simulation models are appropriately applied to detailed operational studies. Like other subarea models, micro simulation models use input from the RTM in order to function in a forecasting mode. They may also be used in conjunction with other larger subarea models. More detailed information on micro simulation models is included in Appendix 2. Current Barriers in the Modelling Area The City of Calgary has had some form of subarea traffic assignment modelling for over ten years. It has, however, been difficult to allocate the level of resources necessary to maintain both the RTM and the sub area models. Since subarea models depend on information from RTM in order to forecast, a choice cannot be made to use subarea models and abandon the RTM. Setting up (calibrating and validating) a subarea model is time consuming and requires a lot of data. Because the models function at a detailed operational level, small deficiencies can cause large errors in the output. Additionally, to be useful for operational studies, subarea models must be kept up-to-date. The amount of time required to establish and maintain subarea models has posed a barrier to their effective application and has lead to criticism of the tool. One of the primary reasons for the difficulty and time required to set up subarea models is the time and detail gap between the current RTM, the subarea models.
Considerable manual effort is required to bridge that gap. The time gap exists because the current RTM is a.m. peak hour only while subarea models are often applied to projects that are focusing on problems that occur in the p.m., or both, peak hours. The RTM can only indirectly inform the development of a p.m. peak hour subarea model. The detail gap occurs because the RTM aggregates data at a relatively course scale to cover the entire city, while a subarea model deals with details in a small area. Under current conditions, the problems encountered with time and resources to establish and maintain CONTRAM subarea models will likely also occur for micro simulation models. Some changes that are underway, as outlined in the section on Changes in Progress, will help alleviate this problem. Changes in Progress Changes to the RTM Work is currently underway to update and expand the RTM. The base travel behaviour data for the current RTM was collected in 1991. At that time there was a pressing need to be ready to respond to GoPlan. There were time and money constraints on model development. The issue of the day was river crossings and the state of the art in modelling, at the time, was peak hour models. The constraints and issues lead to a decision to develop an a.m. peak hour model. In 1993 the basic model was upgraded to incorporate a transit mode choice model. In 1996 a further upgrade was done to include a vehicle occupancy choice model. The 1991 RTM was revalidated (new land use information input and checked against ground counts) in 1994, 1996, and 1999. The RTM that is currently in development will include several new features as follows: a.m., p.m. and off-peak with the capability to simulate peak spreading; more trip purposes; more modes of travel; commercial travel (goods and services trucks); economics of travel; vehicle emissions; the region around Calgary; a more detailed representation of the land use pattern and the road network; and land use / transport interaction (in a second phase of model development). Changes in Micro Simulation Several micro simulation models have been on the market for a few years now. Many more are still in the research and development stage. The majority of micro simulation models do not have the capability to do traffic assignment. In other words, the traffic flows being simulated have to stay on the routes that the operator puts them on. The models were not intelligent enough to allow traffic to change route in response to congestion or network changes. Only a handful of the software packages have this
route choice capability. The addition of route choice capabilities to micro simulation models significantly increases their usefulness. What do the Changes Mean? Firstly, the availability of route choice capability in micro simulation models means that these models are much smarter than they used to be and can now do a similar job to that of the previous Continuous Traffic Assignment Models (CONTRAM). There now exists some overlap between micro simulation and assignment models. However, micro simulation models have the additional, attractive feature of the animated output. It is likely that micro simulation models will replace CONTRAM software for detailed analysis at The City of Calgary over the next couple of years. Secondly, the improvements to the RTM will greatly reduce the gap between the RTM and the subarea models (of al kinds). By adding the p.m. peak hour and increasing the detail in the RTM, it should be possible to reduce the time to calibrate subarea models. The gap should be narrowed considerably with the first stage of the RTM development in 2002 and reduced even further with the second stage of development in 2003-2004.