Verkeer in een landelijk gebied : waarnemingen en analyse van het verkeer in zuidwest Friesland en ontwikkeling van een verkeersmodel

C.F. Jaarsma

    Research output: Thesisinternal PhD, WU

    Abstract

    Transport, the transfer of people or goods, lays at the basis of our welfare. Traffic is the collectivity of moving means of transport. Road traffic especially has grown explosively during the last decades (fig. 1.1). In certain cases these effects (safety; noise; visual annoyance and environment) are radical and inconvenient. Difficulties caused by the extending traffic were revealed at first in the cities and on the main roads, but problems arose in the rural areas also.

    The study described in this dissertation is concerned with traffic in rural areas. The South-West of the province of Friesland has been chosen as the study area. Because road networks should be considered as a connected system, the research is not restricted to rural roads (the agricultural road network) but is also considering the main road network in the rural area (rural highways).

    In chapter 2 extension and development of traffic since 1945 is the main item. Attention is paid to traffic research and community concerns with expanding road traffic. Section 2.2 deals with The Netherlands as a whole; in section 2.3 special attention is given to the rural areas.

    On the Dutch road network there has been a considerable growth in the volume of motorvehicle traffic: roughly speaking it doubled between 1960 and 1970. It grew by more than 50% during the period from 1970 to 1980. Since 1980 it more or less stabilized (fig. 2.2.1).

    Within the category of motorvehicles growth is greatest for cars, which has greatly changed traffic composition (table 2.2.5).

    At first the reaction of our society to growing mobility was positive. A drastic expansion of the surfaced road network was effected, particularly since the early sixties. In the "Tweede Nota over de Ruimtelijke Ordening" (Second Report on Physical Planning), published in 1966, a rectangular network of main roads was designed (total length 5300 km), whereby larger cities are surrounded by one or two networks. The necessity of a more fundamental (model)study of traffic volumes became evident in those years.

    In the years afterwards more attention was paid to the harmful effects of especially mobility by car, on the environment and livability of inner cities. In 1972 the results of "COVW" (Committee Promotion Public Transport in Western Netherlands) and "NEI-Studie" (Study of the Dutch Economic Research Institute) became available. The results of these studies indicate radical consequences of traffic growth if governmental policy remains unchanged. Because these consequences are considered undesirable the predominantly "following policy" regarding development of traffic has gradually been replaced by a "steering policy" ever since 1979.

    The actual long term policy for transport and traffic is described in the "SVV" (Structure Scheme for Traffic and Transport), in which an outline is given of the network of main roads. The total length amounts to about 3400 km, of which 2460 km was realised in 1979. The elaboration of this policy on the short term' takes place in the "MPP" (4-year Program for Traffic and Transport).

    Data on the development of the traffic volume in rural areas are scarce. On the so called 28-points census of the "Landinrichtingsdienst" (Governmental Service of Land Use Planning, abbreviation LD) respectively 294, 406 and 560 personal car units were observed in the years 1962, 1967 and 1972. The number of bicycles and mopeds on these observation points remains nearly constant, but from point to point fluctuation is considerable. The increasing volume is caused by an increasing number of cars. This also appears from an analysis of traffic composition (table 2.3.1.3). These developments are caused by the increase in size and the mechanisation of agricultural holdings, the multiple use of the rural area for outdoor recreation and the increase in the use of minor roads to avoid possible delay on major roads (the so called rat-run traffic).

    The governmental policy with respect to rural roads was published in the "Landbouwwegennota" (Report for Agricultural Roads) in 1954, in the "Plattelandswegennota" (Report for Rural Roads) in 1969 and in the "SLI" (Structure Scheme for Land Use Planning) in 1981. In the SLI the objectives of road development in the rural areas are given. Examples of these objectives are an appropriate and safe access related to the different functions of the rural area, and an expansion of the opportunities for multiple use of land.

    At first the research for rural traffic was almost exclusively directed towards the so called internal farm traffic, of which traffic with farm equipment is the most important feature.

    An extension of the research into external farm traffic is evident. In this traffic lorries, often bulk- or milktank lorries, take an important place.

    In some areas recreational traffic, especially cars, can be very extensive also needs further research.

    In The Netherlands nearly all rural roads are used by all forms of traffic. Therefore the internal and external farm traffic and recreational traffic are part of the total rural traffic. The most important problems concern traffic with farm equipment (moving characteristics), traffic with lorries (road construction, design of roads) and traffic with cars (width of the road). These problems should be solved on a regional basis. Therefore a more precise insight into the present and future traffic volumes and their influencing factors is needed. Transportation models for calculation of link loads up to now hardly applied for rural traffic - seem to be useful.

    With reference to the foregoing, the study described in this dissertation was undertaken, with the following three interconnected objectives:
    - to obtain both data and understanding in character and volume of rural traffic;
    - to obtain insight into this character and volume influencing factors;
    - to describe the relationship between volume and influencing factors by means of a mathematical transportation model.
    These objectives are elaborated in chapter 3.

    Chapter 4 deals with the collecting of data and the first elaboration of these data. First the choice is explained of the area for the study of the character and volume of traffic on main and minor roads in a rural area. South-West Friesland, the rural area South of Workum and West of Lemmer, is chosen as the study area (fig. 4.2.1). Important characteristics of this area are:
    - a predominantly rural area, with a few large lakes and some twenty communities, varying from a few hundreds to a few thousands inhabitants;
    - except for farms, few isolated buildings are to be found;
    - a relatively sparse population, with about 16,000 inhabitants on a land area of about 20,000 hectares;
    - mainly agricultural land use (dairy cattle), but also, of old, an important function for recreation (lakes and wooded areas);
    - a relatively sparse but qualitively good road network, of about 12 m/hectare surfaced non-urban roads.

    Section 4.2 deals with a number of these characteristics (tables 4.2.1/7). Attention is paid to developments in the years 1950-1980.

    To obtain insight into the volume of traffic in an area many "instruments" are available. In section 4.3 is explained that for the study in South-West Friesland the following four methods are used:
    - mechanical counts;
    - visual counts;
    - roadside interviews;
    - home and firm questionnaires.

    In section 4.4 for each of the methods is given: the place where and the time during which the collection of data is done; the object of the inventory; how the fieldwork is done and the results - detailed in chapter 5 - as computed from these data.

    The mechanical counts on the roads are realised from 1972 to 1976 on 30 sites. On nearly all sites during one-and-a-half to two-and-a-half years, information is collected (table 4.4.2.1). On 12 sites both daily volumes and hourly volumes were determined.

    In 1973 on the roads both visual counts and roadside interviews were conducted. These two methods could only be made for a random sample (table 4.4.3.1).

    The home and firm questionnaires are spread over the year 1973, on 25 weekdays and 10 days in the weekend, and directed at inhabitants and firms in 6 parts of the study area, the so called cordons (fig. 4.2.2; table 4.4.5.3.1).

    Section 4.4.6 deals with the question of which relations exist between the different methods. Because the results of the methods differ in time, place and mode only a small overlap occurs. The visual counts and the roadside interviews are mainly used as information extra to the mechanical counts. The home and firm questionnaires do not relate very closely to the other methods, but give insight into character and volume of traffic, and its influencing factors, as demanded in the objectives of this study.

    Chapter 5 deals with traffic characteristics. These are defined in section 5.1 as the ordered, quantitative description of traffic per unit of time. Each characteristic is approached in two ways, namely traffic as produced (or attracted) by inhabitants, and traffic as it appears on the roads. Both approaches show characteristics of the first and the second order. The first order concerns the total amount of trips, the second order the characteristics of trips. This is summarized in the next scheme:

    The summarizing conclusions concerning the traffic characteristics are given in the sections 5.2.4 (for the inhabitants) and 5.3.4 (for the roads).

    It has to be taken into account that traffic is characterised by great fluctuations from day to day. This becomes more evident when the number of "producents" of traffic is smaller. This demonstrates itself especially in data of the home and firm questionnaires, which can only be realised on a restricted number of days, with only a restricted number of households being interviewed each day, because of the manpower available.

    The traffic characteristics of the inhabitants of South-West Friesland are described in section 5.2. The tripproduction, i.e. the average daily trip rate on weekdays, is 2.80, of which 0.94 is by foot, 1.01 by moped and bicycle, 0.51 as driver of a motorvehicle and 0.34 as passenger. In the sections 5.2.2.2.2/14 the relationship between different influencing characteristics (like sex and age, income, etc.) and the tripproduction is presented and discussed.

    Between the cordons the largest differences appear for tripproduction by foot, with extremes of 0.21 and 1.21 (table 5.2.2.2.15.1). This can be explained by differences in geographical situation. Average daily trip rates for vehicle trips vary from 1.47 to 2.39. This variation can be explained by over- or underrepresentation of social classes with relatively high or low tripproductions. The proportional use of buses and trains depends considerable on their availability within a certain distance.

    The average tripproduction for the area as a whole varies from 2.64 (Mondays) to 3.08 (Tuesdays). This result differs between the cordons: different weekdays are found to have the highest or the lowest traffic volumes of the week. For Saturdays and Sundays the average tripproduction for the area is 1.63 and 1.80 (fig. 5.2.2.3.1); large differences occur here also between the cordons.

    The average tripproduction during the months of the year (fig. 5.2.2.4.1) shows a maximum in June (3.49), followed by a minimum in July (2.45). The cordons differ in this respect too. The results can be influenced by error, due to relatively small numbers of observations per month.

    The traffic during the day shows three peak hours, of which the morning peak in the 9th hour is the highest (fig. 5.2.2.6.1). The day pattern differs between the cordons in percentages of traffic per hour. Peak hours, however, are similar for the different cordons.

    On average in South-West Friesland about a quarter of the trips are made for 11 work" and "school" (table 5.2.3.3.1). The distribution by purpose varies widely per cordon. In the "agricultural areas" the proportion of "work traffic" along the roads is relatively low. The proportion of shopping traffic depends a great deal on the geographical situation.

    On an average for South-West Friesland, about 60% of the linked trips occur completely within the cordon where the interviewee is living (table 5.2.3.4.1). Large differences in the geographical trip character between the cordons are found (fig. 5.2.3.4.1), as a result of the differences in the available provisions and/or the geographical situation.

    More than half of the trips are shorter than 1 kilometre. The average triplength measured in a bee-line is 4.3 km (table 5.2.3.5.1). This result varies with the cordon, partly resulting from the composition of the tripproduction by mode.

    Mean trip-duration is 13 minutes (table 5.2.3.6.1). Differences between the cordons can reasonably well be explained by differences in the geographical situation.

    Tripproduction with a vehicle in South-West Friesland is low in comparison with the results of other Dutch studies. This can be explained by the less dominant commuter traffic, by the presence of many relatively older people, by an average lower income and by a low car ownership. The proportion of public transport in tripproduction in South-West Friesland is lower than in large cities, but higher than in medium sized towns. It is above the national average. School traffic has an important place in South-West Friesland, and is characterised by a relatively high proportion of public transport. The mean travelling time in South-West Friesland is 34 minutes per person per day; this is only half of the 67 minutes which was found to be the national average in 1975.

    The traffic characteristics for the roads (c.q. sites) are described in section 5.3. The great variation in volume is remarkable: the AADT varies in 1973 from 167 to 2606 (table 5.3.2.2.1). On an average per year, the busiest site on weekdays is 14 times as busy as the quietest site, on Saturdays 19 times and on Sundays 17 times. On all sites the successive weekdays differ from the weekday-average (AADTw) by at most 10%: the extremes are 90 and 108%. The year-averages for the Saturdays on the different sites show greater differences, with extremes of 77 and 134%. For the Sundays the differences are greatest: 78 and 202% (table 5.3.2.3.1).

    July is on weekdays, for all sites, the month with the highest average volume. This volume however, expressed in percentages of the AADTw, varies greatly from site to site. This is also true for the average volume in the quietest month. Lowest average volume per month appears in different months.

    From figure 5.3.2.5.1 it is clear that, for the roads, different curves for the highest daily volumes are found. Figure 5.3.2.7.1 shows that the same applies for hourly volumes.

    The traffic volumes during the day are characterised by a growth from the 6th until the 12th hour. After a slight decrease in the 13th hour a further increase appears until the peak hour is reached, mostly in the 17th hour (fig. 5.3.2.6.1). On the basis of the year-average the sites can be divided in two groups. The daily peak percentage (expressed in percentages of the daily volume) varies more on the quiet sites. This percentage is in this group on a somewhat higher level than on the busy sites.

    The traffic composition is a traffic characteristic which differs considerable from site to site. The proportion of cars varies on 25 sites on weekdays from 59 to 95%, the proportion of bicycles from 5 to 49% and the proportion of mopeds from 4 to 16% (expressed in percentages of the total number of motorvehicles).

    A great variation can also be found for trip purposes (fig. 5.3.3.3.1), especially for the proportion of recreational traffic. This varies from 9 to 53%, with an average of 34%. For private as well as for business traffic an average proportion is found of one-third (table 5.3.3.3.1).

    The results of the geographical trip character related to South-West Friesland and to the surrounding cordon vary from site tot site, because these results depend greatly upon the situation of the site (figures 5.3.3.4.1/2).

    The trip distances on the roads relate strongly with the trip purposes at the same place (table 5.3.3.5.1). The median values for private, business and recreational traffic are respectively 10, 9 and 19 km.

    For the car occupancy,i.e. the number of persons per car, as well, there is a strong relationship with the trip purposes (fig. 5.3.3.7.1/2). The number of persons per car varies from 1.70 to 3.22, the weekday average is 2.39.

    The sphere of influence for driving for pleasure differs with the site of the interview (fig. 5.3.3.8.1). The sphere of influence of the marked tourist route ("Friese Merenroute"), however, is on different roads always in the same order: it varies from 70 to 100 km (fig. 5.3.3.8.2).

    The average traffic volumes on the roads in South-West Friesland are lower than elsewhere in The Netherlands on comparable roads. Comparing the literature in South-West Friesland a stronger seasonal fluctuation is found. A relation with the recreational function of the area seems logical. This is confirmed by an increasing volume at the weekend, especially on Sundays, on most sites. The rural roads in South-West Friesland are characterised by a low proportion of slow traffic, especially in the summer months and mainly on Sundays. For the trip purposes lower percentages of commuter and business traffic are found, and a relatively high percentage of recreational traffic.

    After the description of the results per characteristic, a comparison is made between the results of the home and firm questionnaire by the inhabitants and the results of the mechanical counts on the roads (section 5.4). It has to be concluded that the comparison is a very difficult one, because of two factors. On one hand a great part of the linked trips of the inhabitants occur within the cordons, so that no mechanical counting site will be passed. On the other hand, three quarters of the traffic on the mechanical counting sites around the cordons is through traffic, without any relation to the inhabitants of the cordons.

    For the traffic volumes by day of the week (table 5.4.2.3.1), and by month of the year (table 5.4.2.4.1), differences appear between the "inhabitant pattern" and the "road pattern". It seems to be logical that the difference is caused by recreants not included in the home and firm questionnaire.

    The traffic volumes during the hours of the day show also differences between both patterns. It seems logical that this is caused by business traffic, which is only found in the mechanical counts on the roads but not in the home and firm questionnaire by the inhabitants. During the afternoon the recreational traffic can also be an important source of differences.

    Chapter 6 deals with the development of a transportation model, by which link loads (daily volumes) for a road network in a rural area can be calculated from explanatory variables (e.g. inhabitants and jobs). The first steps for the development of a model for rural traffic are derived from the type of models as practised for urban areas and main roads. This study deals with a model for the number of motorvehicles on weekdays, whereby the main item of the mechanical counts, the AADTw, is calculated. The transportation model is divided into four phases:

    The successive phases are the determination of the road network, of the trip generation, of the trip distribution and of the assignment, in a sequential model.

    In constructing the model, part of the data, collected in 1973, are used for the calibration of the phases of the model, and another part of the data for testing the model. This is explained in the next scheme:

    Because the calculations in the transportation model are based on traffic attracting and producing zones, first the area has to be zoned. Therefore a grid-system with variable areas is chosen, in which every village with the surrounding rural area forms a separate zone. This produces 28 zones and four external zones (fig. 6.3.1).

    The road network is described by a network consisting of links limited by nodes. This means a simplification with respect to the real network (fig. 6.3.2). The distance of the links is expressed in minutes of travelling time, calculated from the length of the link and the estimated travelling speed. A number of nodes functions as a centroid, hence all trips that start and finish in the belonging zone, are put in the road network.

    In the road network, the routes and the distances (taken together in the term paths) are calculated between the zones. Restriction of the model to 0-paths (shortest paths) does not suffice in a rural road network, where between many couples of zones more routes of similar distance can be chosen. Therefore first and second paths must be calculated as well. This calculation is restricted to paths which are at maximum 35% longer than the 0-path, while a specific node may appear only once in the same path. The calculation time and the computer memory needed for this procedure is considerable (table 6.3.4).

    In the scope of a transportation model tripproduction is defined as the number of departures per zone per day. The motives for this choice are given in section 6.4.3.6. The complement of the tripproduction is the tripattraction, i.e. the number of arrivals per zone per day. Calculated for a day tripproduction and tripattraction are equal. Tripproduction and tripattraction are taken together in the term trip generation. Only cordon passing trips on weekdays, made as driver of a motorvehicle, are taken into account.

    The relation between the trip generation and the explanatory variables is given in mathematical formulae. These are qua techniques based on regression-analysis (formula 6.4.2.1b). The data of the trip generation and the explanatory variables from the cordons are available as input for the regression-analysis. Depending on whether the trip generation of a zone, or the number of trips per household, or per person, is the dependent variable, the terms regression on a zonal basis, on household basis or on personal basis are used. Regression on household or on personal basis are recommended in the literature, because regression models of the zonal type depend strongly on the form and structure of the zones, and thus of the division in zones. Besides, the variation within the zones (which is not taken into account in a zonal model) is much greater than the variation between the zones (table 6.4.2.2).

    Explanatory variables are selected from land use and population characteristics. Explanatory variables with an acceptable value for R-square are not available inside the group of population characteristics, neither for the regression model on household basis (table 6.4.3.3.1), nor for that on personal basis (table 6.4.3.3.2). Preference is given to the number of inhabitants (I) as an explanatory variable for the number of arrivals (A) and departures (V) per zone:
    V = A = 0.474 I (formula 6.4.3.6.1).

    This is confirmed with respect to the acceptable values for R-square and for the F-test (table 6.4.3.2.2). This implies that a choice is made for regression on zonal basis and for an explanatory variable from the group of land use.

    Because of the volume of recreational traffic in the study area, this traffic has to be introduced in the trip generation model. This implies that the number of arrivals on an average weekday has to be estimated for those recreational objects with considerable impact on road traffic volume (table 6.4.4.1). It can be concluded from the summarizing table 6.4.4.2 that for nearly all zones a special contribution of recreation to the trip generation has to be taken in account. The number of arrivals of recreational traffic is on an average one-fifth of the number "normal" arrivals, calculated by formula (6.4.3.6.1). In order to make a more reliable introduction of recreational traffic three extra zones are added.

    In the trip distribution model the number of trips between each two zones is calculated (the so called relation pattern) on the basis of the trip generation determined during the foregoing phase.
    For the distribution the gravity model is used. The basis formula is (6.5.2.1b)

    In order to achieve equality between the number of trips distributed from a zone and its tripproduction, and those distributed to a zone and its tripattraction, two so called constraints are added to the gravity model (formulae 6.5.2.5/9). The coefficients are determined by iteration (formulae 6.5.2.10/11). The effects of the iteration are illustrated in table 6.5.2.1.

    Attention is paid to the distribution function in the gravity model in section 6.5.3. After the analysis of the exponential function, this one is rejected. In the final model the power function (formula 6.5.3.2) is included.

    The unknown parameter in the power function is estimated in such a way that the 0-D-table calculated by the model coincides as accurately as possible with the 0-D-table observed for the cordons with the home and firm questionnaire. The results of the regression calculations (R-square and the coefficient b1, see table 6.5.4.1 and figures 6.5.4.3/4) show a moderate agreement between both 0-D-tables. More or less the same results are found for a wide range (2.0-2.5) of values for the parameter. On this basis the value 2.3 is chosen for the parameter of the power function. For the sake of completeness, a calibration is executed, not including the recreational traffic in the trip generation model. As may be expected, the agreement between the observations and the calculations is less satisfactory (figures 6.5.4.1/2).

    The assignment of the calculated volumes on the road network is the latest phase in the construction of the transportation model. This involves the determination of the routes in the road network, along which the trips, as calculated in the distribution, are made. The assignment results in the calculation of link loads. In this phase, no special observation data are available for calibration. Therefore a choice is made on the basis of study of literature. The most simple assignment model is based on the "all-or-nothing" method, whereby all traffic is assigned to the shortest route. The Kirchhoff-analogy (formula 6.6.2.1) is thought to be the most realistic method for the assignment if more than one route is available. After the completion of the calculations of the assignment phase, it appeared that for nearly two-thirds of the existing relations all traffic is assigned to the shortest path (table 6.6.3.1).

    In section 6.7 the results of the complete transportation model for South- West Friesland are presented. The test with the observations during 1973 is the main item of this section. The most important test is the comparison of the mechanical counts with the calculated link loads (numbers of motorvehicles and vehicle kilometres per weekday). In addition for a number of roads relation patterns and cumulative triplength patterns are also utilised for testing.

    On average the link loads calculated in the model agree to a considerable extent with the mechanical counts:

    There is some overestimation by the model of busy roads and some underestimation of quiet roads (figure and table 6.7.1).

    For roads where a roadside interview was held, the relation pattern of the adjacent cordon as determined by the roadside interview is compared with the relation pattern for that road calculated by the transportation model (table 6.7.3). When the number of relations is not too small, the agreement is often very good; although on some roads, for some relations large differences between model and interview occur. For the sites on the boundary of the study area, the agreement is also reasonably good (table 6.7.4).

    For a number of roads, the cumulative trip-length pattern as calculated with the model can be compared with the cumulative trip-length pattern of the roadside interview (fig. 6.7.2). Due to the calculation method of the model, this comparison has to be restricted to trips inside the study area. Therefore comparison is not possible for trips longer than 15 to 20 km, and the comparison cannot be executed for sites on the boundary of the study area. When these restrictions are taken into account, the agreement is often reasonably good.

    In chapter 7 some items are further commented on, in the form of remarks regarding possible additions or improvements to the reported study.

    Concerning the mechanical counts it is recommended that self-registering counters are used as much as possible, using modified equipment.

    Concerning the home and firm questionnaire it is recommended that the number of addresses per day be increased and that the possession of a drivers licence is registered.

    A tentative study of the development of the traffic volume on three sites in South-West Friesland, shows an average traffic growth of 44% during the period 1973-1983. The growth has been greater on weekdays than on weekend days. For the development of the traffic pattern of the inhabitants, no observations for South-West Friesland are available. Data on Dutch scale give no reason to believe that drastic changes have occurred since 1973 in South-West Friesland.

    Concerning the transportation model, four recommendations are made. More observation data should be made available for the zonal regression-analysis for the trip generation model. More attention should be given to the determination of the relation pattern, which is used for the calibration in the distribution model. Bicycle traffic deserves a place in the model. A better basis for the recreational traffic in the trip generation model has to be developed.

    Original languageDutch
    QualificationDoctor of Philosophy
    Awarding Institution
    Supervisors/Advisors
    • Hellinga, F., Promotor, External person
    Award date14 Dec 1984
    Place of PublicationWageningen
    Publisher
    Publication statusPublished - 1984

    Keywords

    • counting
    • density
    • models
    • road transport
    • traffic
    • transport
    • travel
    • netherlands
    • friesland

    Cite this