Road curve superelevation design: current practices and proposed approach

27 Road curve superelevation design: current practices and proposed approach G. Kanellaidis Abstract Horizontal alignment design standards in the United States and Australia have two basic common features: firstly the absence of a single nationwide maximum superelevation rate and secondly designers' freedom in applying above-minimum values for curve radii. Taking into account the proven dependence of operating speed on curve radius Australian standards introduce the concept of speed environment (characterising highway sections as a whole) to be used alongside the traditional design speed concept (corresponding to individual curves) and incorporate consistency checks as a feedback loop in the design process. Neither of these safeguards is explicitly included in American guidelines. Still in both countries a variety of maximum-superelevation values is used meaning that identically-designed curves (having equal radius and superelevation values) may have resulted from application of different design speeds. In this paper a proposal for simplifying the relationship between radius and superelevation is applied to Australian guidelines for speed environments ranging between 60 km/h and 120 km/h. Consistent application of this proposal for which specification of nationwide maximum superelevation rates is a precondition would result in curve radius serving the driver both as a guide for selecting speed and as a signal for the centrifugal acceleration to be expected thus enhancing horizontal alignment consistency. Refereed Paper This paper has been critically reviewed by at least two recognised experts in the field. Originally submitted: December 1998

28 INTRODUCTION Horizontal alignment design policy for rural roads has traditionally been based on the concept of design speed. However interesting differences are noted in the way that concept is selected and applied in various countries. In several national practices the need has been recognised for design speeds to be more directly based upon actual speed behaviour as well as for checking alignment design on the basis of estimated operating speeds (ERSF 1996; Krammes and Garnham 1995). The basic principle of horizontal curve design is derived from application of the kinematics equation according to which the total lateral acceleration applied through pavement superelevation and tyre pavement friction on a vehicle negotiating a circular curve should be equal to the centrifugal acceleration (CA) due to vehicle movement: CA = _1( = (e +f )g where V is vehicle speed (in m/s) R is curve radius (in m) e is pavement superelevation rate f is the tyre pavement side-friction factor and g is acceleration due to gravity (in m/s2). Application of the design speed concept involves the assumption of a design speed (Vd). For that speed a corresponding side-friction factor value (fvd) is also specified; fvd is commonly a decreasing function of Vd. The designer may select from different possible pairs of values for R and e that satisfy Equation (1) subject to a number of constraints. The most important constraint is that superelevation should fall within a minimum/maximum range the maximum of which (emax) could reflect the risk of stationary vehicles sliding on icy or frozen pavement surfaces. Thus the minimum radius (Rmin) is calculated as follows: Vd = 127 Rmin (emax fvd (2) where Vd is in kilometres per hour (km/h) and R.in is in metres (m). [The numerical coefficient results from conversion between different speed units and multiplication by the value for acceleration due to gravity: (3.6)2 x 9.81-127.] Designers are generally allowed the freedom of applying flatter above-minimum radii for the same design speeds. From Equations (1) and (2) it follows that the sum of (e + f) is reduced proportionally to the increase of R. To achieve that reduction it is common to both apply lower values for e and assume lower values for f. However there is a limiting (minimum) value for the superelevation rate (emin) equal to the rate applied (for the purpose of drainage) in tangent sections. It is noted that above a threshold radius value cross-slope is designed in the same way as for tangent cross-sections. Typically this is the so-called crown section which for these very flat curves provides negative ('adverse') superelevation equalling emin for vehicles moving on the outside of a curve. Diverse procedures are applied internationally regarding not only the design speed concept but also: maximum superelevation rates (1) values of f in relation to the design speed and application of e and f values in curves of aboveminimum radius. The following sections illustrate similarities and differences in the design approaches of the United States (AASHTO 1994) and Australia (Austroads 1993) as regards the above elements of horizontal curve design. Based on empirical findings research suggestions have been made for possible improvements to existing practices. This paper presents a proposal for curve superelevation design on the basis of a simple and consistent relationship between curvature and superelevation. The paper concludes with a summary of main points and identification of issues for further research. DISCUSSION OF EXISTING HORIZONTAL ALIGNMENT DESIGN PRACTICES IN THE UNITED STATES AND AUSTRALIA Design speed The United States (AASHTO 1994) follows what can be called a 'classical implementation' of the design speed concept (Krammes and Garnham 1995). Design speed (Vd) is chosen based on road classification land use and terrain and it is presumed that Vd will not be exceeded.

29 Australian guidelines (Austroads 1993) in contrast define the following four 'speed parameters': (1) desired speed defined as the free speed likely to be adopted by drivers on tangents and other less-constrained elements; (2) speed environment which is numerically equal to the desired speed of the 85th percentile driver and is used to characterise a full road section; (3) design speed applying to individual geometric elements; and (4) limiting curve speed standard defined as the speed beyond which f will exceed its design value (for a given design speed). Especially for speed environments of less than 100 km/h the Australian guidelines recognise that individual curve geometry is determined by but also helps to determine the 85th percentile speed. Thus consistency must be ensured at an early stage by using trial alignments/iterations. Maximum superelevation Both in US (AASHTO 1994) and Australian guidelines (Austroads 1993) a range of possible values for the maximum superelevation rate (emax) is foreseen. This practice can be attributed to the federal structure of both countries (whereby statelevel authorities have the liberty of setting their own limits) as well as their expansion over a large geographical area with varied climatic conditions. The latter is an important factor in defining emax values: in the case of wet or icy pavement conditions superelevation rates of 0.10 or above may be unsafe for slow-moving or still-standing vehicles (especially when tyre quality is poor). In the US the commonest emax values for interurban links are 0.06 0.08 and 0.10; sometimes the more extreme values of 0.12 (provided that snow and ice do not exist) or 0.04 (preferred in urban design) are used. In Australian guidelines recommended emax values range from 0.10 (or sometimes 0.12) in mountainous terrain to 0.06 to 0.07 in flat terrain. The observed variability in emax values in both countries despite having a justification based on the climatic variability (different probability of ice or snow) does not help achieve nationwide consistency in the countries concerned. It can be proven for example that identical curves may have resulted from the application of US guidelines for different design speeds depending on whether the assumed maximum superelevation is 0.06 0.08 or 0.10 (Hayward 1980; Kanellaidis 1991; Krammes 1994). Overall higher emax values (0.10 and above) may be inadvisable even for areas with a low probability of wet or icy pavement conditions since excessive superelevation may mean that drivers of slow moving vehicles (e.g. trucks) can be subjected to negative side friction causing them to steer in the opposite direction to that of the curve. Such situations are undesirable and potentially hazardous for the drivers involved. Side-friction factor in relation to the design speed In the US the design values for side-friction factor (f) have been determined on the basis of experiments using a 'ball-bank indicator' conducted in the 1930s and 1940s in which the main criterion is 'the point at which the centrifugal force is sufficient to cause feelings of discomfort to drivers' (AASHTO 1994). According to the AASHTO guidelines the f is a decreasing function of the design speed changing linearly from 0.16 for 50 km/h to 0.14 for 80 km/h and then linear again but with a steeper slope to 0.10 for 110 km/h and 0.087 for 120 km/h. In Australian guidelines (Austroads 1993) the design values for f have been derived from observations of driver speed behaviour on rural road curves. For speeds above 90 km/h design values are in excess of those likely to be required by the 85th percentile driver. The function of design f against design speed in Australia follows an inverse S-shape; it changes linearly from 0.35 (50 km/h) to 0.31 (70 km/h) falling steeply to 0.12 for 100 km/h and 110 km/h (and to 0.11 for 120 km/h and 130 km/h). The main hazards from vehicle movement on a curve at an excessive speed are skidding and rollover. For the majority of vehicles the critical f values for skidding and rollover far exceed the design values assumed by highway design guidelines (Harwood and Mason 1994) with the possible exception of trucks where instability can occur at values near the Australian guidelines' f value for 50 km/h (0.35). The assumed f values in US and Australian guidelines are decreasing functions of design speed. Empirical data reveal thatf values 'acceptable' by drivers are also higher for lower speeds (McLean 1981; Krammes 1994).

30 Furthermore empirical evidence suggests that the f values accepted by drivers are well in excess of the assumed values especially for lower speeds (below 90 kph) (McLean 1981; Lamm et al. 1989; Krammes 1994). It may be thus concluded that assumed f values are at least at lower speeds conservative in comparison to the f values acceptable by the 85th percentile driver (f85). Table 1 presents (for the speed range of 60 km/h to 120 km/h) a comparison of assumed f values in US and Australian guidelines. Superelevation and side-friction values in curves of above-minimum radius The procedure described in United States guidelines (AASHTO 1994) recognises the possibility that curves with above-minimum radius will be overdriven (driven at speeds above the design speed) but makes no corrections to the design speed. Rather it provides an increasing ratio of e over f for increasing curve radius resulting in a parabolic relationship between e and the inverse of R (1/R). The limiting value emirs is equal to 0.02. In Australian guidelines (Austroads 1993) it is only mentioned that when above-minimum radii are selected the corresponding superelevation and side friction values are 'below their maximum values'. No exact calculations are made and no special graphs are developed for above-minimum radii. However the limiting curve speed standard is for given e and R values the speed at which the corresponding design value off is reached thus indirectly providing a lower limit. Minimum superelevation values range from 0.02 to 0.03 the latter value being typical for bituminous pavements. Discussion on the effect of curvature and superelevation on driving behaviour There is ample and consistent empirical evidence linking operating speed to radius of curvature. Relationships have been developed over the years showing that the inverse of curve radius 1/R can be a good predictor of speed (Taragin 1954; Lamm et al. 1989; Kanellaidis et al. 1990; Ottesen and Krammes 1994). In current US design practice (AASHTO 1994) it is usually assumed that the design speed is unchanged that is unaffected by changes in curvature. Thus curves of an above-minimum radius only serve to reduce the total centrifugal acceleration requirement (e + f) which is directly proportional to degree of curve (or inversely proportional to curve radius). However this consideration does not take into account the fact that operating speed is affected by the horizontal alignment. In certain cases this omission may lead to a serious underestimation of operating speed. This question is addressed in a different way in Australian guidelines (Austroads 1993) where design speed does not correspond to a whole section but to individual elements. Thus given a speed environment design speed for a horizontal curve is Table 1 Assumed side-friction factors (f) in design guidelines of the United States and Australia for speeds between 60 km/h and 120 km/h Design speed Vd (km/h) AASHTO (United States) Assumed side friction factors (f ) Austroads (Australia) 60 0.153 0.33 70 0.146 0.31 80 0.140 0.26 90 0.127 0.18 100 0.113 0.12 110 0.100 0.12 120 0.087 0.11

31 a function of that speed environment and the curve radius as shown in Figure 2.2 of the Australian guidelines. With the help of an iterative process (using trial alignments subject to consistency checking) the Australian guidelines provide the potential for abetter matching between design speed and actual operating velocities. However in both countries given regional differences in maximum superelevation rates it is possible that identical curves (that is having equal radii and superelevation rates) correspond to different design speeds. A large diversity of superelevation rates and the lack of a clear association between superelevation and curvature have been observed (Krammes 1994). Also there is empirical evidence that superelevation does not have a significant influence on driving behaviour parameters such as operating speed (Gambard and Louah 1986) or acceptable f values (Kanellaidis and Dimitropoulos 1995). Thus although the dynamics of vehicle movement show that the selection of superelevation is important for traffic safety research findings suggest that it does not make much of a difference for drivers who are primarily affected by the radius of curvature in choosing their speed. Therefore there appears to be a case for highway design practice to provide a clearer link between speed curvature and superelevation; this would further refine the link between curvature and speed environment that is already inherent in Australian practice. If a given curve radius corresponded to a certain (and appropriately selected) superelevation rate then the speed chosen because of that radius would result in a specific (and acceptable) level of overall lateral acceleration (e + f). Thus the radius would serve the driver not only as a guide as to the speed to be chosen but also as a signal for the centrifugal acceleration to be expected. In this manner the consistency of the alignment could be further enhanced. PROPOSAL FOR IMPROVING CURVE SUPERELEVATION DESIGN PRACTICE Compared to other countries' practices the Australian horizontal alignment design procedure (Austroads 1993) follows a considerably advanced approach in dealing with the issue of providing a proper matching between assumed and actual speeds. The selection of design speed is not made through what has elsewhere been criticised as an 'arbitrary' process. On the contrary the influence of alignment design (and especially curve radius) on operating speeds is recognised and the Australian definition of design speed corresponds to individual curves (as contrasted to 'speed environment' which characterises a whole highway section). Through an iterative procedure alignment is fine-tuned so as to help attain both consistency of successive curves and harmonisation between assumed and operating speeds. The concept of using an assumed speed for the whole section together with assumed speeds for individual elements the latter being more closely related to expected actual speeds is applied in a similar form in Italy (Krammes and Garnham 1995) where the 'range' of design speeds is the equivalent of the speed environment. Kanellaidis and Dimitropoulos (1995) have mentioned a similar twostage concept for curve design consisting of design speeds (defining the minimum radius) and 'speed standards' (corresponding to a range of radius values); speed standards may be higher (but not lower) than the design speed reflecting the fact that at these flatter curves operating speed may exceed the design speed. In addition the importance of horizontal alignment consistency (given the effect of horizontal curvature on operating speeds) should not be underestimated (Leisch and Leisch 1977; Messer et al. 1981; Lamm et al. 1992; Krammes 1994). Large operating-speed differentials are known to be correlated to increased accident occurrence (Anderson and Krammes 1994). Australian guidelines are among those incorporating research proposals for design standards to include consistency checks as a 'feedback loop' (Lamm et al. 1992; Kanellaidis 1996). If the designers' liberty of applying above-minimum radius values is to be preserved (and indeed it may be unnecessarily restrictive to do away with aboveminimum design altogether) it is important to ensure that above-minimum design does not lead to potential safety problems. Flatter curves are associated with higher operating speeds and due to conservative assumptions for the design f design speed is (even at minimum-radius curves especially in US standards) an underestimation of operating speed. Therefore design guidelines should be enhanced with additional consistency safeguards regarding above-minimum design.

32 One important inconsistency factor is the variety of emax values applicable in the practices of Australia the United States and other countries. It has been argued that agreement on a single emax value possibly at the rate of 0.08 (assuming high-type pavements) would be beneficial and feasible contributing to greater consistency (Craus and Livneh 1978; Kanellaidis 1991; Krammes 1994). In addition simplification of the radius superelevation relationship so that there is a one-toone correspondence between R and e could also lead to greater consistency as already discussed. The idea of simplifying the R e relationship is not new. There have been arguments and actual proposals for a change in that direction (Koeppel 1986; Kanellaidis 1991; Nicholson 1998). This paper presents the possibility of applying such a proposal defined originally by Kanellaidis and Dimitropoulos (1995) to the design of two-lane rural roads in Australia. In applying this proposal it is assumed that the designer follows the recommendations of Australian guidelines regarding the selection of speed environment. Within the selected speed environment design speed for individual curves is defined in association with the chosen radius (Austroads 1993). The proposal's innovation lies in the introduction for each speed environment of a one-to-one correspondence between radius and superelevation. The proposal is based on four criteria. The first two are given in Australian design guidelines whereas the additional criteria correspond to additional safety considerations suggested by research. Criterion 1: Superelevation should be between emin and emax For Australian nationwide consistency it is proposed that these values are set at 0.02 and 0.08 respectively. Criterion 2: The side friction factor should not exceed the specified maximum value fmax' corresponding to the speed environment as defined in Australian standards (see Austroads values in Table 1). Criterion 3: The portion of the total sideways force that is provided by superelevation that is the ratio of e over (e + f) should be higher than a desirable minimum value to provide a satisfactory safety margin. It has been suggested that this ratio should be at least 0.25 (Craus and Livneh 1978; Kanellaidis and Dimitropoulos 1995). Criterion 4: The 'hands-off' speed defined as the threshold speed between positive and negative side friction is an important parameter regarding the safety and comfort of slower drivers. British guidelines (Highway Link Design 1984) require that the hands-off speed should be at most equal to the predicted 15th percentile free speed; the ratio of that speed to the design speed is estimated at around 0.60. The limiting condition for this criterion is specified by applying a superelevation rate such that the ratio of [ e / (e + f) ] equals 0.36 since (e + f) is proportional to the square power of design speed (Vd)2. Therefore the criterion is satisfied for [ e / (e + f) < 0.36. If the above criteria are applied for a certain speed environment (which determines the maximum f value) then in the corresponding radius superelevation graph (R e graph) there is an area within which all four criteria are satisfied. Within this area it is possible to define a simple R e relationship by which each allowable radius value corresponds to one single superelevation rate. The paper's proposal specifies possible R e relationships for speed environments between 60 km/h and 120 km/h (in increments of 10 km/h). Figure 1 Figure 2 Figure 3 and Figure 4 illustrate the application of the four criteria as well as the proposed relationships for the speed environments of 60 km/ h 80 km/h 100 km/h and 120 km/h. The proposed relationships consist of linear segments linking rounded pairs of values (usually with an accuracy of 50 or 100 metres for radius and 0.01 for superelevation) intended to provide a practical proposal. Figure 5 presents an overview of the proposal for the range of speed environments examined. It is noted that for speed environments of 100 km/h and above there are no pairs of R e values with e > 0.06 satisfying all four criteria. Therefore for those speed environments the upper parts of the graphs (shown in dotted lines) are replaced by the limiting values of Criterion 2 (maximum f value); for these values Criterion 4 (maximum 'hands-off speed') is not satisfied. It may thus be recommended to avoid superelevation rates exceeding 0.06 for speed

33 0.08 0.07. Superelevation rate (m /m) 0.06 0.05 0.04 Criterion 2 Criterion 4 ; Criterion 3 s s...... 0.03. -.. 0.02. 0 100 200 300 Curve radius (m) 400 500 600 Figure 1 Proposed R-e relationship for the speed environment of 60 km/h 0.08 II 1 k Superelevation rate (m/m) 0.07 0.06 0.05 0.04 0.03 + Criterion 4 i ' 4 ` II Criterion 3 s s Criterion 4 N. It....... ---- ---- ---- 0.02 0 100 200 300 400 500 600 Curve radius (m) 700 800 900 1000 Figure 2 Proposed R-e relationship for the speed environment of 80 km/h

34 0.08 0.07 Superelevation rate (m/m) 0.06 0.05 0.04 Cnterion 4 0.03 0.02 0 100 200 300 Criterion 2 Cnterion 3 400 500 600 700 800 900 1000 1100 1200 1300 1400 1500 Curve radius (m) Figure 3 Proposed R-e relationship for the speed environment of 100 km/h 0.08 0.07 Superelevation rate (m/m) 0.06 0.05 0.04.Criterion 4 0.03 Criterion 2 Criterion 3 0.02 0 100 200 300 400 500 600 700 800 900 1000 1100 1200 13(X) 1403 1500 1630 11(i) 1833 1933 2330 2100 Curve radius (m) Figure 4 Proposed R-e relationship for the speed environment of 120 km/h

007 1.* 006 E re 008 o 005 ca as c. 004 003 002 89 121158 246 60 70 394 476 597 MI = i t I t i I t II I I I I I I I I I I I I I t I I 80 510 _aa 90 100 694 11 907 120 1148 141 1714 0 100 200 300 400 500 600 700 800 900 1000 1100 1200 1300 1400 1500 1600 1700 1800 1900 2000 2100 2200 Figure 5 Proposed R-e relationships for speed environments between 60 km/h and 120 km/h Curve Radius (m) 2040. Road curve superelevation design: current practices and proposed approach

36 environments of 100 km/h and above in order to provide increased comfort and safety for the slowermoving traffic. SUMMARY AND FURTHER ISSUES The Australian highway design guidelines (Austroads 1993) include certain provisions for improving the safety of horizontal curve design. In contrast to US guidelines (AASHTO 1994) where the traditional design-speed approach is utilised Australian standards include four distinct speed parameters: desired speed speed environment design speed and limiting curve speed standard. Thus the two-way interaction between curve geometry and operating speed is acknowledged and trial alignments are used in order to ensure the highest possible degree of horizontal alignment consistency. However the absence of a nationwide maximum superelevation value in combination with designers' freedom (within the consistency constraints specified) to use 'above-minimum' curve radii with a corresponding reduction in superelevation rate and side friction factor may lead to cases where identical curves (with equal radius and superelevation values) correspond to different assumed speeds. Since the curve radius has been proven to be the key determinant of actual operating speeds it is advisable to design curves in such a way that curve radius can be a signal for the total centrifugal acceleration to be expected by the 85th percentile driver; this canbe achieved by more closely linking curvature and superelevation in design practice. This paper has presented a proposal for narrowing down the theoretically infinite choices of aboveminimum curve radii and corresponding superelevation rates. To that end the following criteria are used: (1) a nationwide range of possible superelevation rates between 0.02 and 0.08; (2) a maximum side friction factor depending on the speed environment as specified in Australian guidelines; (3) superelevation rate such as to counter at least one-quarter of the total centrifugal acceleration (safety margin); (4) 'hands-off speed' not exceeding 60 per cent of the speed environment value (provision for the safety and comfort of slower drivers). Criteria (3) and (4) serve to introduce additional safety elements in curve superelevation design going one step further than the Australian guidelines' already substantial existing safeguards of minimum design values and consistency checks. Application of the criteria can be the first step towards applying one-to-one relationships between radius and superelevation for various speed environments. This paper's proposal uses pairs of rounded values leading to simple and straightforward relationships. It also ensures with the exception of a few cases corresponding to maximum or minimum superelevation rates that a specific pair of radius superelevation values corresponds to one single speed environment. The proposed approach is a step towards further enhancing horizontal alignment consistency and as such can be generally beneficial to driver comfort and safety. Driving behaviour variability implies of course that perception of the 'feedback' from the design would be stronger for some drivers and subtler for others. The above criteria or nationally-defined variants taking into account the different national approaches could be adapted to the horizontal curve design procedures of other countries too. At a next stage the approach could be extended to also cover very flat curves (determination of the threshold for removing adverse superelevation). One possible area for further improvement that is not addressed by this paper's proposal concerns the fact that f values acceptable by the 85th percentile driver tend to exceed the design values (especially for speed environments below 90 km/h). Since the application of higher superelevation rates (0.10 or greater) is generally not recommended the question to be posed is whether it would be feasible to raise the design values for f in order to achieve a better matching between design and operating speeds. Considering that the vector of tyre pavement friction on horizontal curves has a tangential and a lateral component any increase in the allowable side friction value (lateral component) will mean a decrease in the available tangential friction coefficient. Therefore if side friction factors were to be increased the consequences with respect to stopping (that is a

37 reduction of available tangential friction factor leading to an increase in required stopping sight distance) should be taken into account and weighed against whatever benefits would arise from matching design speed and operating speed. Since tyre pavement friction inventories often reveal large disparities the choice of f values is by necessity conservative; in the longer term improved quality and consistency in both pavement and tyre quality may lead to increased design f values which may correspond more closely to acceptablef values than is the case today. REFERENCES AASHTO (1994). A Policy on Geometric Design of Highways and Streets American Association of State Highway and Transportation Officials (AASHTO) Washington D.C. ANDERSON I.B. and KRAMMES R.A. (1994). Speed Eduction as a Surrogate for Accident Experience at Horizontal Curves on Rural Two-Lane Highways 73rd Annual Meeting of the Transportation Research Board Washington D.C. AUSTROADS (1993). Rural Road Design Guide to the Geometric Design of Rural Roads Austroads Sydney. CRAUS J. and LIVNEH M. (1978). Superelevation and Curvature of Horizontal Curves Transportation Research Record 685 pp. 7-13 TRB National Research Council Washington D.C. ERSF (EUROPEAN ROAD SAFETY FEDERATION) (1996). Intersafe Technical Guide on Road Safety forinterurban Roads ERSF Brussels. GAMBARD J.M. and LOUAH G. (1986). Free Speed as a Function of Road Geometrical Characteristics 14th PTRC Annual Meeting Brighton United Kingdom. HARWOOD D.W. and MASON J.M. Jr. (1994). Horizontal Curve Design for Passenger Cars and Trucks 73rd Annual Meeting of the Transportation Research Board Washington D.C. HAYWARD J. (1980). Highway Alignment and Superelevation: Some Design Speed Misconceptions Transportation Research Record 757 pp. 22-5 TRB National Research Council Washington D.C. HIGHWAY LINK DESIGN (1984) Departmental Advice Note TA 43/84 Department of Transport London United Kingdom. KANELLAIDIS G. (1991). Aspects of Highway Superelevation Design Journal oftransportation Engineering Vol. 117 No. 6 pp. 624-32 American Society of Civil Engineers. KANELLAIDIS G. (1996). Human Factors in Highway Geometric Design Journal of Transportation Engineering Vol. 122 No. 1 pp. 59-66 American Society of Civil Engineers. KANELLAIDIS G. and DIMITROPOULOS I. (1995). Investigation of Current and Proposed Superelevation Design Practices on Roadway Curves TRB International Symposium on Highway Geometric Design Practices Boston Mass. KANELLAIDIS G. GOLIAS J. and EFSTATHIADIS S. (1990). Drivers' Speed Behaviour on Rural Road Curves Traffic Engineering and Control Vol. 31 No. 7 pp. 414-15. KOEPPEL G. (1986). Strassenentwurf Rueckblick und Ausblick Strasse und Autobahn No. 9 pp. 395-402 Federal Republic of Germany (in German). KRAMMES R.A. (1994). Design Speed and Operating Speed in Rural Highway Alignment Design 73rd Annual Meeting of the Transportation Research Board Washington D.C. KRAMMES R.A. and GARNHAM M.A. (1995). Review of Alignment Design Policies Worldwide TRB International Symposium on Highway Geometric Design Practices Boston Mass. LAMM R. CHOUEIRI E. and MAILAENDER T. (1989). Side Friction Demand Versus Side Friction Assumed for Curve Design on Two-Lane Rural Highways Transportation Research Record 1303 pp. 11-21 TRB National Research Council Washington D.C. LAMM R. GUENTHER A. K. and CHOUEIRI E.M. (1992). Safety Module for Highway Design: Applied Manually or Using CAD 71st Annual Meeting of the Transportation Research Board Washington D.C. LEISCH J.E. and LEISCH J.P. (1977). New Concepts in Design Speed Application Transportation Research Record 631 pp. 4-14 TRB National Research Council Washington D.C. McLEAN J.R. (1981). Driver Speed Behaviour and Rural Road Alignment Design Traffic Engineering and Control No. 4 pp. 208-11. MESSER C.J. MOUNCE J.M. and BRACKETT R.Q. (1981). Highway Geometric Design Consistency Related to Driver Expectancy Vols. II and III Federal Highway Administration Washington D.C. NICHOLSON A. (1998). Superelevation Side Friction and Roadway Consistency Journal of Transportation Engineering Vol. 124 No. 5 pp. 411-18 American Society of Civil Engineers. OTTESEN J.L. and KRAMMES R.A. (1994). Speed Profile Model for a US Operating-Speed-Based Design Consistency Evaluation Procedure 73rd Annual Meeting of the Transportation Research Board Washington D.C. TARAGIN A. (1954). Driver Performance on Horizontal Curves Proceedings of the Thirty-Third Annual Meeting of the Highway Research Board pp. 446-66 TRB National Research Council Washington D.C.

38 George Kanellaidis is a Civil and Transportation Engineer with more than 25 years of expertise in a large number of Greek and international projects and research in the field of highway engineering human factors and road safety. His current position is Associate Professor in the Department of Transportation Planning and Engineering at the National Technical University of Athens. Contact Dr George Kanellaidis National Technical University of Athens Department of Transportation Planning and Engineering Iroon Polytechniou 5 15773 Zografou / Athens Greece Tel +30 1 772 1283 Fax +30 1 772 1327 E-mail g-kanel@central.ntua.gr