Circular and Spiral Curve Ranging: Five Worked Examples
Simple and transition curve computations worked step by step from field data: curve insertion at a PI, the effect of a radius revision, and spiral geometry in both degrees and grads.
These examples come from a field diary kept on a route-surveying assignment between 3 and 18 October 2003, worked by hand at the time in a mixture of degrees and grads. Every figure below has been recomputed from the original given data before publication. Where the recomputation disagreed with what was written in 2003, the corrected value is shown and the disagreement is explained rather than quietly fixed.
Conventions throughout: coordinates are (Northing, Easting) in metres, bearings are whole-circle bearings measured clockwise from north, and chainages use station notation, so 10+500 means 10,500 m along the line.
1. Inserting a simple curve at a point of intersection
A route runs through three successive intersection points. A circular curve of radius 300 m is to be inserted at the middle one.
| Point | Northing (m) | Easting (m) | Note |
|---|---|---|---|
| PI-35 | 97315.000 | 31396.000 | |
| PI-36 | 98068.000 | 32169.000 | Station 10+500.00 |
| PI-37 | 98248.000 | 32946.000 |
The deflection angle is not given. It has to come from the coordinates, which is the part most people get wrong when they start on curves: the angle you need is the difference between the two tangent bearings, not an angle measured in the field.
Step 1, the two tangent bearings.
PI-35 to PI-36: dN = 753.000 dE = 773.000
D = 1079.138 m
brg = atan(773.000 / 753.000) = 45° 45' 03.18"
PI-36 to PI-37: dN = 180.000 dE = 777.000
D = 797.577 m
brg = atan(777.000 / 180.000) = 76° 57' 24.92"
Step 2, the deflection angle.
Δ = 76° 57' 24.92" − 45° 45' 03.18" = 31° 12' 21.74"
Step 3, tangent length and curve length.
T = R tan(Δ/2) = 300 × tan(15° 36' 10.87") = 83.779 m
L = πRΔ/180 = π × 300 × 31.20604 / 180 = 163.394 m
Step 4, chainages.
BC = 10+500.000 − 83.779 = 10+416.221
EC = 10+416.221 + 163.394 = 10+579.616
Step 5, coordinates of the curve ends. BC lies a distance T back from the PI along the reverse of the back bearing, and EC lies a distance T forward along the onward bearing.
BC: N = 98068.000 + 83.779 cos(225° 45' 03.18") = 98009.541
E = 32169.000 + 83.779 sin(225° 45' 03.18") = 32108.988
EC: N = 98068.000 + 83.779 cos(76° 57' 24.92") = 98086.907
E = 32169.000 + 83.779 sin(76° 57' 24.92") = 32250.617
The 2003 notebook recorded EC as 10+579.615. Carried at full precision it is 10+579.616. One millimetre on a curve end is of no practical consequence, but it is worth knowing where it comes from: the hand calculation used π as 3.1415, and that truncation shows up in the fourth significant figure of L.
Everything else in this problem reproduces exactly, including both sets of coordinates to the millimetre.
2. What happens when the radius is revised
This one is more useful than it looks, because design revisions late in a project are normal and the station bookkeeping catches people out.
A curve is originally designed with R = 1000 m at PI station 5+318, deflection angle 25.6210 grads. The radius is then reduced to 900 m, with the PI unchanged.
Original T = 1000 tan(25.6210ᵍ/2) = 203.988 m
L = π × 1000 × 25.6210/200 = 402.454 m
BC = 5+114.012 EC = 5+516.466
Revised T = 900 tan(25.6210ᵍ/2) = 183.589 m
L = π × 900 × 25.6210/200 = 362.208 m
BC = 5+134.411 EC = 5+496.619
A correction to the original notes. The 2003 entry concluded that the new BC and EC are “each 20.399 m closer to the PI”. That is true of the BC and not of the EC, and the difference matters if you are re-pegging from existing marks.
The BC moves by the change in tangent length alone:
BC shift = ΔT = (1000 − 900) tan(25.6210ᵍ/2) = 20.399 m
The EC moves by the change in tangent length and the change in curve length, which pull in opposite directions:
EC = PI − T + L
EC shift = ΔL − ΔT = 40.246 − 20.399 = 19.847 m
So the BC comes forward 20.399 m and the EC comes back 19.847 m, a difference of 552 mm. The original note went on to add 20.399 m to the new EC and said this “essentially recovers” the old EC station. It lands on 5+517.018 against a true 5+516.466, and the half-metre gap is exactly the discrepancy above rather than rounding.
If you take one thing from this example, take that: on a fixed PI, a radius change does not slide the whole curve rigidly. Both ends move toward the PI, but by different amounts.
3. Spiral transitions, and what a longer spiral costs you
A spiral-circle-spiral curve connects the tangents through three PIs with R = 450 m and spiral parameter A = 200. The question asked at the time was whether increasing A to 250, with the radius held, lengthens or shortens the circular arc.
Deflection from coordinates: Δ = 19° 43' 51.70"
A = 200: Ls = A²/R = 200²/450 = 88.889 m
θs = Ls/(2R) = 5° 39' 31.83"
Δc = Δ − 2θs = 8° 24' 48.04"
Lc = R Δc = 66.078 m
A = 250: Ls = 250²/450 = 138.889 m
θs = 8° 50' 30.99"
Δc = 2° 02' 49.73"
Lc = 16.078 m
The arc shrinks from 66.078 m to 16.078 m, and the reduction of exactly 50.000 m is not a coincidence. Substituting the spiral angle into the arc length collapses the whole thing:
Lc = R(Δ − 2θs) = R(Δ − Ls/R) = RΔ − Ls
The circular arc is the full curve length minus one spiral length. Hold R and Δ fixed, and every metre you add to the spiral comes straight off the arc, one for one. Here Ls grew by 50.000 m, so the arc lost 50.000 m.
That relation is worth carrying in your head. It tells you immediately when a spiral has become too long for the deflection available, because the arc goes to zero at Ls = RΔ, and anything beyond that is not a buildable spiral-circle-spiral at all.
4. Main tangent and external distance
Full spiral geometry for R = 800 m, A = 400, Δ = 36.6320 grads.
Ls = A²/R = 400²/800 = 200.000 m
θs = Ls/(2R) = 7.9577ᵍ
ΔR = Ls²/(24R) = 200²/(24 × 800) = 2.083 m (shift)
Xs = Ls[1 − Ls²/(40R²)] = 199.688 m (long tangent)
The main tangent and external distance then follow:
M.T. = (R + ΔR) tan(Δ/2) + Xs − R sin θs
= 802.083 × tan(18.3160ᵍ) + 199.688 − 800 × sin(7.9577ᵍ)
= 237.350 + 199.688 − 99.740
= 337.298 m
E.S. = (R + ΔR)/cos(Δ/2) − R
= 802.083 / cos(18.3160ᵍ) − 800
= 36.465 m
Two corrections here. The compiled notes printed the first term as 436.994 and the third as 99.384. Both are wrong: they are 237.350 and 99.740. You can tell the first one is wrong without computing anything, because the three terms as printed sum to 537.298 rather than the 337.298 the same page reports as its answer.
The final M.T. and E.S. are both right, so the error was in transcribing the working rather than in the calculation. It is a good argument for always writing the sum out and checking that the parts add up to the total you quoted.
5. A complete spiral element set
R = 500 m, A = 300, total deflection 60.0000 grads.
Ls = 300²/500 = 180.000 m
θs = Ls × 100/(πR) = 180 × 100/(π × 500) = 11.4592ᵍ
Δc = 60.0000 − 2(11.4592) = 37.0817ᵍ
Lc = π × 500 × 37.0817/200 = 291.239 m
Check it with the shortcut from example 3:
Lc = RΔ − Ls = 500 × (60ᵍ in radians) − 180 = 471.239 − 180 = 291.239 m
Both routes agree, which is the point of knowing the second one.
Working in grads
Three of these five examples are in grads, because that was the convention on the instrument in use. The conversions are simple enough that there is no excuse for getting them wrong, but mixing systems mid-calculation is a common and expensive mistake:
| Quantity | Degrees | Grads |
|---|---|---|
| Full circle | 360° | 400ᵍ |
| Right angle | 90° | 100ᵍ |
| Curve length | L = πRΔ/180 | L = πRΔ/200 |
| Spiral angle | θs = Ls·90/(πR) | θs = Ls·100/(πR) |
| Conversion | 1ᵍ = 0.9° | 1° = 1.1111ᵍ |
The one that bites is the curve length denominator. Using 180 with an angle in grads overstates the curve by about 11 per cent, which is large enough to be obvious on a long curve and small enough to survive a careless check on a short one.
What the recomputation found
Fifty of the fifty-seven figures checked across the whole set reproduce from the given data to within half a millimetre. Of the rest, one was a genuine propagating arithmetic error in a traverse, two were wrong intermediates printed in the spiral working above, one was the conceptual slip about BC and EC shifts, and three were sub-millimetre rounding.
The methods and the engineering conclusions in the 2003 notes hold up. The arithmetic mostly does too. The places it does not are the ordinary ones: truncating π, carrying rounded intermediates forward, and copying a line of working across wrongly.
