You are on page 1of 16

James A.

Rosenow 1

Performance-Based Superelevation Transition Design


March 20, 2015
Word Count: 7,495 (5,995 in text; 1,500 equivalent in figures)

Corresponding Author:
James A. Rosenow
Minnesota Department of Transportation
395 John Ireland Boulevard
Mail Stop 686
St. Paul, MN 55155
Phone: (651) 366-4673
Fax: (651) 366-4680
james.rosenow@state.mn.us
James A. Rosenow 2

1 ABSTRACT
2 Design criteria for highway superelevation transition rate and distribution with respect to spiral
3 curve alignment points date from the publication Transition Curves for Highways by Joseph
4 Barnett (1940). AASHTO criteria governing superelevation transition distribution with respect to
5 simple curve points have not been revised since their first publication by AASHO in the 1950s.
6 These sources encourage transition rates slower than maxima that are based on both comfort and
7 appearance. The distribution recommended by Barnett superimposes the runoff over the length
8 of the spiral, placing the level point at the TS/ST. For simple curves, AASHTO advises placing
9 60 to 90 percent of the runoff length on the tangent based on the premise that the natural
10 transitional path in the absence of a spiral extends disproportionately into the tangent.
11 Anecdotal experience has cast doubt on the soundness of these criteria. Driver accounts
12 of negative side friction approaching simple curves suggest excessive superelevation being
13 developed well prior to the actual need to counteract developing curvature. To investigate
14 mathematically, a point-mass force balance was computed in spreadsheet form to calculate side
15 friction through the tangent-to-curve transition. The outcome confirms negative side friction
16 occurring in the outside lane using standard transition rates and distribution proportions but more
17 comfortable operation using non-standard distributions and rates. Moreover, it indicates that
18 transition parameters are best tailored to specific combinations of design speed, curvature and
19 superelevation rate. The conclusion: Unique transition designs should be determined for
20 individual curves in just such an analytical fashion, necessitating more flexible standard policies.
James A. Rosenow 3

1 INTRODUCTION
2 Every superelevated highway curve necessitates a transition from the normally crowned cross
3 section on tangent to the fully superelevated section within the curve. It is comprised of the
4 tangent runout the length of roadway through which the outside lane transitions from normal
5 cross slope to zero (flat) and the superelevation runoff, through which the lane transitions from
6 zero to full superelevation. The length of this transition as well as its longitudinal placement is
7 governed by design criteria that are based on appearance, driver comfort and safety.
8 Design criteria for the placement of superelevation transitions with respect to spiral
9 transition elements and/or simple curve points have been largely unchanged since the publication
10 of the earliest texts governing highway engineering. Current guides and their historical forebears
11 alternately make reference to empirical methods and theoretical considerations as bases for
12 their policy statements, although little or no actual mathematical evidence is provided therein. (1,
13 2, 3, 4, 5, 6, 7, 8, 9) In general, these policies admittedly err on the side of earlier introduction of
14 superelevation in the tangent-to-curve transition into a simple curve, i.e. a large proportion of the
15 superelevation runoff on the tangent prior to the horizontal point of curvature (P.C.) so as to
16 diminish the side friction demand at any point in the transition.
17 Anecdotal user experience calls this practice into question. Early development of
18 superelevation on the tangent prior to the curve has been observed to result in negative side
19 friction, a condition in which there is insufficient centripetal acceleration to counteract gravity.
20 As a result, drivers must steer uphill against the developing superelevation, opposite the
21 direction of the ensuing curve. The potential for this phenomenon was acknowledged in early
22 AASHO publications on the subject but was evidently considered acceptable as long as
23 excessive positive side friction is not incurred. (2, 3, 4, 5, 6) Very recent research on the
24 dynamics of superelevation has once again identified the phenomenon of negative side friction as
25 indeed occurring in tangent-to-curve transitions having standard transition placements. (10)
26 In all ensuing discussion, the design condition will be understood to be the transition
27 from tangent into curve, although it is understood that the equivalent but inverse condition will
28 occur in the curve-to-tangent transition.

29 OBJECTIVE
30 The objectives of this investigation are to mathematically verify negative side friction that has
31 been observed occurring in superelevation transitions; identify the circumstances under which
32 negative or markedly varying side friction would be expected to occur; and devise a simple and
33 easily duplicated methodology with which to engineer superelevation transitions attuned to
34 driver comfort, one that can be tailored to specific combinations of design speed, curvature and
35 superelevation rate. Both the placement and the rate of transition are subjects of investigation, as
36 they both stand to affect the side friction experienced by the driver.

37 HISTORY OF DESIGN GUIDANCE


38 In Transition Curves for Highways (1) by Joseph Barnett, the placement of superelevation
39 transitions was discussed with respect to spiral transitions into and out of curves. It stated,
40 Curvature begins at the T.S. [tangent-to-spiral point], at which point the outer half of the
41 pavement should be level so that at no point on the curve will the surface slope downward
42 toward the outside. Full superelevation should be attained at the S.C. [spiral-to-curve point]
43 where the radius of curvature is that for which the superelevation is designed. In other words,
44 the superelevation runoff was prescribed to be placed over the length of the spiral, beginning and
James A. Rosenow 4

1 ending at the T.S. and S.C. respectively. Barnett recommended a maximum transition rate of
2 1:150 (measured as the slope of the outer pavement edge with respect to the roadway centerline)
3 for an assumed design speed of 30 mph; 1:175 for a design speed of 40 mph; and 1:200 for
4 higher speeds.
5 A Policy on Geometric Design of Rural Highways (2, 3) adopted Barnetts criteria for
6 runoff placement with respect to spirals. For transitions into simple curves, it stated, Empirical
7 methods are employed to locate the superelevation runoff length with respect to the P.C. No
8 method for division between the tangent and circular curve can be completely rationalized. No
9 further mention of empirical methods was made, suggesting the criteria that follow were based
10 on observation and/or intuition. It went on to state, In general, theoretical considerations favor
11 the practice of placing a larger proportion of the runoff length on the approach tangent rather
12 than on the circular curve. The resultant superelevation on the tangent is undesirable in that the
13 driver may have to steer in a direction opposite to the direction of the curve ahead to stay in line
14 but the maximum side friction developed, which is equal to the rate of applied superelevation, is
15 at all times below the rate of side friction considered comfortable and safeWhile the side
16 friction factor developed upon the tangent is undesirable, the development on curves of friction
17 factors greatly in excess of the design basis results in hazardous conditions. In these statements,
18 negative side friction due to a high proportion of runoff on tangent was recognized and
19 acknowledged to be undesirable; however, AASHO ordered it as less of a concern than the
20 imposition of excessive side friction, especially that exceeding the maximum rates established
21 elsewhere therein. No basis for the cited theoretical considerations was provided. The Policy
22 then stated that, due to the natural spiral path adopted by drivers into a simple curve, What
23 might appear to be an undesirable cross slope upon the tangent actually compensates for the
24 curvilinear path of the vehicle. And what may be considered lack of superelevation at the
25 beginning of the circular curve proper is compensated for by the vehicle traveling a curvilinear
26 path that is flatter than the roadway circular arcIt is evident from the above that [the placement
27 of the runoff] cannot be determined exactly from available practice and information. In general,
28 design with 50 to 100 percent of the length of superelevation runoff on the tangent can be
29 considered as suitable. For a more precise design control it is concluded that from 60 to 80
30 percent of the length of runoff preferably should be located on the tangent
31 In A Policy on Geometric Design of Highways and Streets (4), little was changed from
32 the rural highway policy that preceded it. However, this first edition of the so-called Green Book
33 did add a remark near the beginning of the discussion on placing the runoff with simple curve
34 geometry, stating, Current design practice is to place approximately two-thirds of the runoff on
35 the tangent approach and one-third on the curve. This presumably reflects widespread practice
36 that had emerged by that time, which itself may have been influenced by the prior AASHO
37 guidance.
38 The second and third editions of the Green Book (5, 6) offered only one subtle but
39 significant modification. Friction factors greatly in excess of the design basis were characterized
40 as a worse condition as opposed to the prior wording of hazardous conditions. This
41 presumably reflects a realization that such conditions could not be positively associated with
42 crash incidence.
43 The fourth through sixth editions of the Green Book (7, 8, 9) feature a substantially re-
44 written discussion reflecting research conducted in this time period, most notably NCHRP
45 Report 439 (10), discussed later herein. AASHTO acknowledges the natural spiral path created
46 by the driver in entering a circular curve as well as its research-established two- to four-second
James A. Rosenow 5

1 length. The Green Book also implies that the natural spiral is balanced evenly across the P.C.
2 when it states, the peak lateral acceleration incurred at the P.C. should theoretically be equal
3 to 50 percent of the lateral acceleration associated with the circular curve. Echoing the findings
4 in Report 439, however, it recommends a vast majority of the transition be placed on the tangent
5 to avoid lateral velocity in an outward direction resulting in a corrective steering maneuver
6 that produces a path radius sharper than that of the roadway curve and consequently an
7 undesirable increase in peak side friction demand and potential lateral shift of the vehicle into
8 an adjacent lane. Table 3-18 in the Green Books sixth edition (9) exhibits recommended runoff
9 placement in accordance with Report 439, stating this practice should minimize lateral
10 acceleration and the vehicles lateral motion. The accompanying text modifies the longstanding
11 60-to-80 percent range to 60 to 90 percent in order to encompass the table values, stating they
12 offer the best operating conditions. The discussion in prior editions of AASHO/AASHTO
13 policy describing undesirable aspects of negative side friction in the transition is absent in the
14 contemporary Policy.

15 PERTINENT RECENT RESEARCH

16 NCHRP Report 439


17 It is beyond this papers scope to offer a critique of NCHRP Report 439 (11) other than
18 incidentally. This subsection will instead serve to cite data and conclusions therein that are
19 useful to this research objective as well as to inform observations and hypotheses presented
20 below.
21 A central component of this study was a kinematic analysis seeking to analyze a vehicles
22 lateral acceleration and associated lateral motion through the tangent-to-curve transition and into
23 the curve proper. A stated objective was to recommend superelevation transition criteria that,
24 when applied, minimize or eliminate lateral shift (drift) with respect to the traffic lane. It
25 identified outward shift as being particularly problematic due to it requiring a corrective steering
26 action that incurs a peak side friction demand exceeding the stated maximum design value. The
27 study did not appear to address the existence, magnitude or consequence of unbalanced force or
28 lateral shift to the inside of the curve. Regardless, the kinematic modeling methodology would
29 seem to be valid and the results meaningful.
30 The authors cite and presumably validate prior study of driver behavior through tangent-
31 to-curve transitions (and whose findings are incorporated into their kinematic model). Those
32 findings are that drivers create a spiral path of a length ranging from 2.0 to 3.1 seconds of travel
33 time and that the spiral path is approximately centered on the P.C.

34 NCHRP Report 774


35 NCHRP Project 15-39 ostensibly studied dynamics associated with combinations of
36 curvature/superelevation and steep downhill grades; in doing so it investigated transitional
37 conditions including tangent to curve. Report 774 (10) found that, in relatively extreme
38 conditions superelevation rates higher than 12% lower friction margins available to the driver
39 were exhibited near the curve P.C. This was due to the amount of superelevation developed on
40 the tangent with such a high maximum rate as well as, in the words of the report, due to the
41 vehicle steering up the superelevated roadway on the approach, then reversing steering inputs
42 toward the inside of the curve. This indicates a dynamic physical aspect to the tangent-to-curve
43 transition and accompanying superelevation transition beyond what would be evident in a
44 simple point-mass calculation.
James A. Rosenow 6

1 In the researchers judgment, this phenomenon is worth considering throughout the range
2 of design conditions, not just with extreme superelevation rates. Report 774 recommends that
3 the superelevation achieved at the P.C. not exceed a threshold value governed by a formula
4 provided therein. It states, If the condition presented above is met, the superelevation transition
5 may be placed as indicated in Green Book Table 3-18. If the condition presented above is not
6 met, designers should reduce the proportion of the maximum superelevation attained at the P.C.
7 of the horizontal curve, or introduce a spiral transition curve between the approach tangent and
8 simple horizontal curve.
9 It must be noted that the above aspect of this analysis evidently does not take into account
10 the natural spiral path created by drivers into a simple curve. Regardless, it should be recognized
11 for its key and heretofore unknown finding: undesirable effects related to negative side friction
12 transitioning into positive friction demand as measured by a sophisticated analysis of vehicle
13 dynamics.

14 OBSERVATION AND HYPOTHESIS


15 As the history of design criteria and recent research provided above demonstrates, the presence
16 of negative side friction or even pronounced variations in side friction through the tangent-to-
17 curve transition has been recognized but not considered a significant factor to be taken into
18 account in policy or practice. Only the most recent research effort NCHRP Project 15-39
19 (Report 774) quantifies these effects on drivers and large vehicles and recommends tailoring
20 designs to avoid it in excess (10). This finding notwithstanding, the general approach advocated
21 by current policy and criteria entails erring on the side of oversupply of superelevation in the
22 transition so as to avoid any condition where the driver might conceivably be subjected to side
23 friction exceeding the maximum allowable values provided in the Green Book. The logic and
24 internal consistency of this approach is questionable in the following aspects:
25 The Green Book does cite negative and varying side friction as its primary factor in not
26 recommending the Method 3 superelevation distribution. It states that, for vehicles
27 traveling at average running speed, this superelevation method results in negative friction
28 for curves from very flat radii to about the middle of the range of curve radii
29 Subjecting the driver to markedly different side friction conditions in different situations
30 is characterized as not logical and may result in erratic driving, either at the design or
31 average running speed (9).
32 The premise of avoiding side friction even marginally exceeding design maxima is not
33 supported by experience or research. These design values are an approximated aggregate
34 of several disparate criteria that each sought to establish a reasonable limit of comfort for
35 open road design and which provide considerable margin of safety against skidding, even
36 under non-ideal road and vehicle conditions. (12) Report 774 confirms that, in all but the
37 lowest speed conditions, the margin between friction supply and demand is large in the
38 range of 0.25 to 0.3 using conservative skidding assumptions (10). Empirically there is
39 no known evidence of performance problems due to slight variations in superelevation
40 rate and, by extension, side friction demand. The AASHTO Highway Safety Manual (13)
41 shows no crash effects on rural two-lane highways where superelevation rates vary by 1.0
42 percent or less from Method 5 design values.
43 AASHO/AASHTO policies and their predecessor publications prescribe placing the
44 runoff over the length of a spiral transition, a criterion that survives to the present day.
45 Considering, though, that the natural spiral path created by the driver into a simple curve
James A. Rosenow 7

1 is approximately centered on the P.C., the equivalent runoff placement onto that simple
2 curve would be 50 percent each on tangent and on curve. This inconsistency in criteria
3 for runoff placement with respect to the driven spiral path is not acknowledged and
4 perhaps not recognized.

5 This paper documents current and past policy in order to provide sufficient background
6 as well as to assert that the matters raised and analyzed herein have likely been overlooked or at
7 least undervalued through history. This investigation represents a fresh analytical approach to
8 the policy questions, in effect providing an independent review of the current design criteria and
9 catalogue of prior analyses.
10 The fundamental hypothesis of this investigation is that driver comfort is in fact not
11 optimized by essentially overdesigning superelevation supply through the tangent-to-curve
12 transition and that this longstanding practice indeed results in driver discomfort and potentially
13 erratic driving. This is based on observations that drivers are experiencing greatly varying side
14 friction conditions and negative side friction, requiring them to apply steering effort back and
15 forth as well as uphill against the developing superelevation in order to maintain position within
16 their lane. It should be possible to confirm these anecdotal observations with a simple physical
17 analysis. Such an analysis should be able to test the validity of current design criteria and/or
18 suggest alternative criteria. The design parameter in greatest question is the Green Book
19 recommendation to place between 60 and 90 percent of runoff on tangent with greater
20 percentages characterized as desirable. Also of interest is the runoff length/rate. Secondarily,
21 this research seeks to establish a simple and easily applicable methodology for designing
22 individual superelevation transitions to optimize the driver comfort aspect of performance,
23 tailored to unique combinations of curvature and speed.

24 INVESTIGATIVE APPROACH
25 The point-mass model is utilized by AASHTO and others to represent the simplified physics of a
26 vehicle traveling around a horizontal curve. It is used herein as a suitably accurate and simple
27 analytical method as well as to provide a familiar frame of reference to established nomenclature
28 and terms. The fundamental formula governing vehicle operation around a curve is
+ 2
29 = Equation 1
1 15

30 where: e = rate of roadway superelevation (unitless)


31 f = side friction (demand) factor (unitless)
32 V = vehicle speed, mph
33 R = radius of curve, ft

34 As the term ef is always small, the formula simplifies to


2
35 + = Equation 2
15

36 and can be rearranged to solve for f, thus:


2
37 = Equation 3
15
James A. Rosenow 8

1 Using this formula, one can solve for the friction demand on the vehicle at any point in
2 the tangent-to-curve transition given the instantaneous curve radius, the vehicles travel speed
3 and the instantaneous superelevation rate. By performing this computation at tight intervals
4 through the transition, a graph of the output data indicates how a drivers steering effort changes
5 through the transition including any necessary steering uphill, counter to the superelevation
6 (i.e. negative values of f).
7 Modeling of simple curves as well as curves with spiral transitions was desired in order
8 to compare the two conditions analytically and to develop a design tool applicable to either
9 geometry. For the simple curve condition, it was necessary to make an assumption regarding the
10 driving path from tangent to curve. Adopting the long-accepted premise discussed herein that a
11 natural spiral path is created, approximately centered on the curve P.C., is logical. The length of
12 the assumed spiral transition can be based on findings also cited above, which indicated a length
13 of 2.0 to 3.1 seconds of travel time. The most commonly accepted value within that range is 2.0
14 seconds, which is presented in the Green Book as a desirable length for a constructed spiral (9).
15 The combination of an assumed spiral location centered on the P.C. and a 2.0-second spiral
16 length is also logical in that it results in mathematically minimizing drivers shift within their
17 lanes and within the bounds of comfort, something drivers are naturally inclined to do.
18 A spreadsheet was created to perform this computation every two feet longitudinally
19 through a tangent-to-curve transition. Separate tabs exist for simple curves and spiral-
20 transitioned curves. In both cases, the spreadsheet computes instantaneous curve radii through
21 the area of the actual or assumed spiral. Within either tab, the operator enters values for travel
22 speed, degree of curvature / curve radius, full superelevation rate and transition rate. In the
23 simple curve tab, the operator enters the percent of the transition on tangent; in the spiral curve
24 tab, the length of spiral is entered. As tested, the spiral tab was hard wired to fix the end of the
25 runoff at the S.C. (i.e. end of spiral) and to place the transition backward from that point. The
26 condition modeled is a simple two-lane configuration, which best describes one direction of a
27 four-lane divided highway. The spreadsheet models the entire transition including
28 superelevation runoff and tangent runout and assumes 2.0 percent normal cross slope, a constant
29 rate of transition (i.e. not changing transition rates within the runoff and/or runout) and a flat
30 plane rotation scheme (as opposed to a rotating crown scheme). As discussed above, the
31 condition modeled and illustrated is the tangent-to-curve transition.
32 Computations were performed for a number of combinations of speed, curve radius and
33 superelevation rate based on common conditions encountered in design. Within each condition,
34 the percentage of transition on tangent and the transition rate were toggled in a trial-and-error
35 manner, and the effects on side friction through the transition were observed.
James A. Rosenow 9

1 RESULTS

2 Simple Curve Analyses


3 The initial conditions modeled were of common and known field conditions under which
4 negative side friction has been experienced. The first was a 2-degree (2,865-foot radius) simple
5 curve with a 70 mph design speed and a full superelevation rate of 0.055 (5.5 percent) from the
6 Green Books Method 5 superelevation distribution (based on a maximum superelevation rate of
7 6.0 percent). The transition parameters are two-thirds of the runoff placed on the tangent and a
8 1:400 rate of transition (relative gradient), both of which are standard practice in Minnesota. The
9 assumed speed is the design speed. The output is shown graphically in Figure 1.

10 FIGURE 1 Trial Output (V=70 mph, R=2,865 ft, e=0.055, S=1:400, 67% runoff on
11 tangent).

12 The inside and outside lanes, respectively, are relative to the curve i.e. inside of the
13 curve and outside of the curve. To visualize the effect of the depicted side friction curves on the
14 driver, they should be thought of as the steering effort required to sustain the vehicle along the
15 prescribed travel path, which in this case includes an assumed spiral transition 205 feet in length
16 (2.0 seconds of travel time). On the Y-axis, positive friction indicates active steering in the
17 direction of the impending curve, while negative friction indicates active steering in the opposite
18 direction of the curve. On the tangent, the side friction demands represent the steering correction
19 against the normal cross slope of the crowned roadway. The output confirms the hypothesis and
20 physical observation that negative side friction does occur in the outside lane for a portion of the
21 transition from tangent to curve, based on the assumptions and parameters stated. Moreover,
22 marked variation in side friction demand occurs for the outside-lane driver: back and forth in
23 one direction, then in the opposite, and finally in the original direction of cross slope correction
24 up to the terminal side friction factor of 0.059. This again appears to support the hypothesis that
25 employing the standard transition parameters can result in significant discomfort for drivers.
James A. Rosenow 10

1 To gauge the effect travel speed has on side friction variation, a trial was run utilizing a
2 running speed of 58 mph, based on the 70 mph design speed and Table 3-6 in the AASHTO
3 Green Book (9). The output is shown in Figure 2. The lower assumed travel speed had the
4 effect of deepening the negative side friction experienced; however the terminal side friction in
5 the curve proper is not as great, so the overall side friction variation for the outside-lane traveler
6 through the transition is not as marked.

7 FIGURE 2 Trial Output (V=58 mph, R=2,865 ft, e=0.055, S=1:400, 67% runoff on
8 tangent).

9 Trial and error was subsequently employed to arrive at a combination of superelevation


10 transition placement and development rate that would be expected to optimize driver comfort in
11 the tangent-to-curve maneuver based on the assumptions stated herein. For these purposes,
12 driver comfort is considered to be a minimum of variation in side friction demand, especially
13 variation that entails reversing side friction from positive to negative or vice versa. Additionally,
14 a condition under which side friction exceeds the terminal value experienced in the curve proper
15 would be considered undesirable and potentially unacceptable if it exceeds the maximum side
16 friction factor for the design speed. Using the 70 mph design speed as the assumed running
17 speed, the exercise yielded the parameters and friction curve in Figure 3.
James A. Rosenow 11

1 FIGURE 3 Trial Output (V=70 mph, R=2,865 ft, e=0.055, S=1:225, 33% runoff on
2 tangent).

3 It is evident from the simple shape of the outside-lane friction curve that this arrangement
4 would stand to optimize the steering maneuver from the standpoint of comfort and ease, at least
5 within the assumptions for driving path stated herein. The inside-lane driver is subjected to a
6 change in the rate of increase in steering effort during the transition, but at no point does a
7 reversal of steering direction take place. (The trial-and-error process seems to indicate that the
8 only way to avoid this rate-of-change adjustment would be to transition the lanes at different
9 rates, which is largely considered impractical and eschews the flat plane rotation scheme used
10 nearly universally in the industry.) The percentage of runoff on tangent 33 percent is notable
11 for its deviation from Green Book policy and long-standing practice of placing no less than 50
12 percent on tangent. Also notable is the seemingly optimal transition rate of 1:225, which is faster
13 than the Green Book maximum of 1:250 for the design speed (9). Although at no point on the
14 curves does the friction demand exceed the terminal value or standard maximum, the trials
15 indicated that the use of a slower transition rate and/or a shorter assumed spiral would cause the
16 curve(s) to spike above the terminal friction factor. This is an illustration of how actual driving
17 behavior is a sizable determining factor in how a curve transition operates. It does not, however,
18 affect the validity of the above finding, as any such driving variation would stand to have the
19 same relative effect on any friction curve resulting from any combination of transition rate and
20 runoff placement. In the final analysis, a transition length/rate tailored to drivers likely spiral
21 path would still appear to provide sound performance outcomes.
22 Particularly prominent in the anecdotal observations of driver discomfort negotiating
23 tangent-to-curve transitions are very wide highways and freeways with multiple lanes in
24 superelevation transition. These circumstances are subject to criteria requiring transition lengths
25 to be increased to avoid problems relating to drainage and driver comfort resulting from steep
26 relative gradients at the roadway edge furthest from the rotation point. Unfortunately this tends
27 to exacerbate side friction variations, as can be seen in Figure 4, which depicts the friction factor
James A. Rosenow 12

1 computation for a curve on an in-place multilane freeway in Minneapolis with a 60 mph design
2 speed, constructed with a 1:750 transition rate and 50 percent of runoff on the tangent.

3 FIGURE 4 Computational Output for I-94 near Broadway Street in Minneapolis


4 (V=60 mph, R=1,763 ft, e=0.056, S=1:750, 50% of runoff on tangent).

5 In this case, due primarily to the length of the transition, side friction in the outside lane
6 becomes negative nearly 250 feet in advance of the curve P.C., and friction demand spikes well
7 above the terminal value of 0.080 (but does not exceed the maximum f of 0.12 for the design
8 speed). Assumption of a longer driver-created spiral path would ease these problems, but
9 adopting such a path would carry a vehicle well into an adjacent lane. Additionally, even
10 constructing a spiraled geometry of this length would be questionable, as research finds that
11 drivers are disinclined to follow an excessively long spiral element (11). Trials indicate that
12 varying the percentage of runoff on tangent serves only to shift the problem longitudinally along
13 the roadway, either increasing the length of the negative side friction region or worsening the
14 severity of the friction spike within the curve.

15 Spiral Curve Analyses


16 The most notable spiral curve condition modeled was the same common condition modeled
17 initially for simple curves a 2-degree curve with 5.5 percent superelevation and a 70 mph
18 design speed but with a 300-foot spiral transition between tangent and curve. The 300-foot
19 spiral length was chosen due to its commonness on the highway system. The initial trial
20 incorporated standard runoff placement, which is fit to the length of the spiral, in this case
21 amounting to a rate of approximately 1:455. The graphical result is presented in Figure 5.
James A. Rosenow 13

1 FIGURE 5 Trial Output (V=70 mph, R=2,865 ft, L s =300, e=0.055, S=1:455, runoff fit to
2 spiral length).

3 Although negative side friction does not result, the outside lane friction demand does dip
4 to zero before increasing to terminal, due to the tangent runout beginning 109 feet in advance of
5 the T.S. As with the initial simple curve analysis, a trial was run using AASHTOs assumed
6 running speed of 58 mph. The characteristic of the output is similar to that in Figure 2 for the
7 simple curve condition (including slightly negative side friction) and is not pictured.
8 Lastly, trial and error was used to arrive at an arrangement that would seem to optimize
9 driver comfort, as was done with simple curves above. The output is shown in Figure 6. The
10 design speed of 70 mph is assumed as the travel speed. The transition rate accomplishing this
11 condition was 1:330. Superelevation runoff ends at the end of the spiral, which follows intuition;
12 likewise, the tangent runout begins at the beginning of the spiral. This would appear to be an
13 arrangement that could be applied universally to optimize comfort for most if not all spiraled
14 curves.
James A. Rosenow 14

1 FIGURE 6 Trial Output (V=70 mph, R=2,865 ft, L s =300, e=0.055, S=1:330, runoff ends at
2 S.C.).

3 CONCLUSIONS AND RECOMMENDATIONS


4 A number of computations have been performed using this methodology since its creation, both
5 for research and policy development as well as in project development to explore opportunities
6 for design flexibility and optimal performance. The trials/computations described above provide
7 representative examples, but the full range of experience is beyond the scope of this paper.
8 Based on this experience, the following conclusions have been reached.
9 1. This methodology confirms observations of negative side friction and markedly varying
10 friction demand in known anecdotal driving experience with simple curves designed
11 using standard parameters for rate of transition and percentage of runoff on the tangent.
12 This supports the hypothesis that these standard criteria do not optimize driver comfort
13 and indeed appear to create uncomfortable tangent-to-curve transitions and potentially
14 erratic driving behavior.
15 2. Erratic driving behavior has been anecdotally observed with these curves, although
16 additional study would be necessary to positively correlate it with particular geometric
17 features.
18 3. Optimal driver comfort into simple curves at least from the standpoint of a simple and
19 straightforward steering maneuver without markedly varying and reversing side friction
20 demands appears to be provided by tailoring transition rate and runoff placement to the
21 specific circumstances. The optimal designs generally feature much less runoff placed on
22 the tangent than recommended by AASHTO in many cases less than 50 percent of
23 runoff on tangent and sometimes exhibit transition rates faster than recommended in
24 AASHTO criteria. Optimal transition rates are generally faster for higher superelevation
25 design values and vice versa.
James A. Rosenow 15

1 4. Optimal driver comfort into spiraled curves appears to be provided by fitting the full
2 transition tangent runout and superelevation runoff to the length of spiral. In other
3 words, the runout would begin at the T.S., and the runoff would end at the S.C.
4 5. Unbalanced side friction early in the spiral or on the tangent in the case of a simple
5 curve is small (no greater than 0.04 at the midpoint of the spiral), not abrupt, and
6 arguably far superior to negative side friction due to oversupply of superelevation. In the
7 case of both a constructed spiral and a driver-created spiral, the introduction of curvature
8 is very gradual, and superelevation is not needed early in its length.
9 6. As evidenced by the multiple-lane example computation, long superelevation transitions
10 lead to apparent discomfort early in the transition and again well into the curve proper.
11 7. The overarching conclusion is that, based on these analyses, driver comfort through
12 tangent-to-curve transitions is not served by the AASHTO recommendation to provide up
13 to 90 percent of the runoff on tangent. This criterion is intended to prevent discomfort
14 and safety deficit associated with greater-than-maximum friction factor demand late in
15 the transition due to over-steering or steering correction, yet there is no known evidence
16 that such driver behavior is common enough to design for or that any such occurrences
17 result in actual performance problems. Instead, the criterion appears to unwittingly create
18 driver discomfort and potential erratic driving early in the transition.
19 8. This methodology represents a straightforward, performance-based and evidently valid
20 way to design superelevation transitions. Consistent with the emerging direction of the
21 road design industry, it utilizes analysis to design for performance and requires design
22 flexibility to allow for optimal solutions.

23 Based on these conclusions and the apparent correlation between the computation outputs and
24 observation, the following recommendations are offered.
25 1. This methodology should be explored further as a new, performance-based standard.
26 Although the referenced spreadsheet tool is crude and somewhat inflexible, more
27 sophisticated tools can be developed, perhaps imbedded in road design software
28 packages.
29 2. Additional research should be considered that would correlate measured effects on the
30 driver and vehicle and erratic driving behavior with these geometric design features.
31 3. Until such time as the above recommendations can be vetted and implemented, additional
32 flexibility should be provided in the AASHTO Green Book pertaining to both percentage
33 of runoff on tangent and transition rate. The recommendation to place the vast majority
34 of runoff on the tangent is highly suspect and should be reconsidered or at least softened.
35 4. The AASHTO Green Book and other applicable guides should discuss the physical
36 concepts investigated herein, specifically the concept of tailoring design parameters to
37 balance physical forces and design for a friction demand outcome, as opposed to the
38 nominal criteria currently featured. This dovetails to some degree with recommendations
39 in NCHRP Report 774 discussed previously. Sufficient technical guidance could be
40 provided for practitioners to compute side friction demand at key points in the transition.
James A. Rosenow 16

1 REFERENCES
2 1. Barnett, J. Transition Curves for Highways. Federal Works Agency, Public Road
3 Administration, Washington, DC, 1940.
4 2. AASHO. A Policy on Geometric Design of Rural Highways. American Association of
5 State Highway Officials, Washington, DC, 1954.
6 3. AASHO. A Policy on Geometric Design of Rural Highways. American Association of
7 State Highway Officials, Washington, DC, 1965.
8 4. AASHTO. A Policy on Geometric Design of Highways and Streets. American
9 Association of State Highway and Transportation Officials, Washington, DC, 1984.
10 5. AASHTO. A Policy on Geometric Design of Highways and Streets, 2nd Edition.
11 American Association of State Highway and Transportation Officials, Washington, DC,
12 1990.
13 6. AASHTO. A Policy on Geometric Design of Highways and Streets, 3rd Edition.
14 American Association of State Highway and Transportation Officials, Washington, DC,
15 1994.
16 7. AASHTO. A Policy on Geometric Design of Highways and Streets, 4th Edition.
17 American Association of State Highway and Transportation Officials, Washington, DC,
18 2001.
19 8. AASHTO. A Policy on Geometric Design of Highways and Streets, 5th Edition.
20 American Association of State Highway and Transportation Officials, Washington, DC,
21 2004.
22 9. AASHTO. A Policy on Geometric Design of Highways and Streets, 6th Edition.
23 American Association of State Highway and Transportation Officials, Washington, DC,
24 2011.
25 10. Torbic, D. J., M. K. OLaughlin, D. W. Harwood, K. M. Bauer, C. D. Bokenkroger, L.
26 M. Lucas, J. R. Ronchetto, S. Brennan, E. Donnell, A. Brown and T. Varunjikar.
27 National Cooperative Highway Research Program Report 774: Superelevation Criteria
28 for Sharp Horizontal Curves on Steep Grades. NCHRP, Transportation Research Board,
29 Washington, DC, 2014.
30 11. Bonneson, J. A. National Cooperative Highway Research Program Report 439:
31 Superelevation Distribution Methods and Transition Designs. NCHRP, Transportation
32 Research Board, Washington, DC, 2000.
33 12. Tan, Carol H. An investigation of comfortable lateral acceleration on horizontal curves.
34 PhD dissertation, Pennsylvania State University, 2005.
35 13. AASHTO. Highway Safety Manual. American Association of State Highway and
36 Transportation Officials, Washington, DC, 2010.

You might also like