Available online at www.sciencedirect.

com

ScienceDirect
Procedia - Social and Behavioral Sciences 96 (2013) 1277 – 1285

13th COTA International Conference of Transportation Professionals (CICTP 2013)

Accuracy Comparison of Roadway Earthwork Computation

between 3D and 2D Methods
Jian-chuan Chenga,b,* , Long-jian Jianga,b
a
School of Transportation, Southeast University, Nanjing 210096, China
b
Jiangsu Provincial University-Enterprise Base for Graduate Education on Roadway Safety Design and Evaluation, Nanjing 210005, China

Abstract

This study aims to calculate the earthwork volume in 3D method that has been seldom used before in roadway engineering,
and to reconfirm the feasibility of average-end-area method for earthwork volume that is widely used in literature. After
reviewing the related studies and comparing various CAD packages, the analysis of accuracy difference between 3D method
and average-end-area method is conducted. It shows that in average-end-area method the critical value of interval distance
between two consecutive cross sections is 30m. It also shows that the Change Rate of Cut-Fill (CRCF) value, an index firstly
proposed to represent the cut-fill variance frequency associated with roadway terrain, alignment and profile design, has no
significant impacts on the accuracy of 2D result. It is concluded that the 3D method could be easily used in practice with the
CAD software. Meanwhile, average-end-area method with less than the critical interval distance between two consecutive
cross sections can guarantee the earthwork calculation accuracy.

©© 2013 TheAuthors.
2013 The Authors. Published
Published by Elsevier
by Elsevier B.V.access under CC BY-NC-ND license.
Ltd. Open
Selection and/or
Selection and peer-review
peer-review under responsibility
under responsibility ofOverseas
of Chinese ChineseTransportation
Overseas Transportation Association (COTA).
Association (COTA).

Keywords:earthwork volume; highway design; 3D method ; average-end-area method

1. Introduction

Reliable and accurate earthwork volume calculation is one of the most important components in roadway
engineering that can influence the choosing of roadway alignment, the cost and construction.
As the appearance and widely application of Digital Terrain Model (DTM), roadway design has stepped into
3D era and accordingly 3D method for earthwork volume calculation is also developed. But the concept of
adopting average-end-area method (hereinafter also as 2D method) is deep-rooted in roadway design. According
to an investigation in US, 87% and 91% designers use average-end-area for design estimates and final quantities

* Corresponding author. Tel.:+86-025-8379-0385; fax: +86-025-8379-5184.

E-mail address: jccheng@seu.edu.cn

1877-0428 © 2013 The Authors. Published by Elsevier Ltd. Open access under CC BY-NC-ND license.
Selection and peer-review under responsibility of Chinese Overseas Transportation Association (COTA).
doi:10.1016/j.sbspro.2013.08.145
1278 Jian-chuan Cheng and Long-jian Jiang / Procedia - Social and Behavioral Sciences 96 (2013) 1277 – 1285

respectively, and 97% of the respondents recognize average-end-area in their policies, standards, and procedural
documents. All these data show 2D method still enjoys popular support (Hintz & Vonderohe [1], 2011).
Obviously the result of average-end-area method is just an approximate value, and theoretically speaking, result
by calculating the volume between existing ground and design surface is the precise value of earthwork-which is
also usually called surface-to-surface method that based on DTM technology. From this point of view, DTM has
its technical advantage in earthwork volume calculation.
Civil 3D software package, developed by Autodesk Corporation in this context, is a Building Information
Modeling (BIM) solution for civil engineering that can be used in transportation, land development and water
conservancy project, supporting 3D volume calculation based on DTM. This study chooses Civil 3D package as
the tool to compute earthwork volume in 3D method.
This paper verifies the feasibility of calculating the earthwork volume in 3D method by Civil 3D software
based on DTM and compares the accuracy difference between 2D and 3D methods for the reference of practical
engineering. Moreover, the study also gives an explanation why inaccurate 2D method actually prevails in
engineering and a critical interval distance to guarantee accuracy in practical use.

2. Literature Review

The earthwork volume is one of the most important objectives in horizontal and vertical alignment
optimization, so most researches firstly focused on the cut-fill balancing to minimize the cost. Stark and Nicholls
(1972) started to employ linear programming into earthwork optimization and this method was developed by
Mayer and Stark (1981) and Nandgaonkar (1981). Easa (1988) [2] integrated the selection of roadway grades and
the minimizing of earthwork into one problem by enumerating all technically feasible grades and solving the
linear programming problem. But Easa s method couldn t guarantee the global optimality, so Moreb (1996) [3]
proposed a model that succeeded not only in reducing the time problem but also arriving at global optimality
based on Easa s theory. Goktepe and Lav (2003) [4] then developed a method called weighted ground
elevation that considered the material properties in grade line selection to balance the cut and fill volume. All
this researches were conducted by average-end-method.
When some researchers realized the imprecision and limitations of 2D method in volume calculation (Easa,
1989 [5] & 1992 [6]), they started to develop some improvements in average-end-method. The prismoidal
formula which can enhance the accuracy is the most familiar improvement to us. Anderson, Hikhail and
Woolnough [7] (1985) presented a Pappus-based model to compute the volume in horizontal curved area. Easa
(1992) [8] developed a mathematical model that calculates earthwork volumes based on triple integration in
horizontal curve, but it was applicable when d between the stations
and the ground cross slope between the stations was linear and constant. So Easa (2003) [9] estimated earthwork
volumes of curved roadways in Monte Carlo simulation method but this model still wasn t extended to the case
of combined roadway alignments (where horizontal and vertical curves overlap). Kim and Schonfeld (2001) [10]
introduced two methods using vector and parametric representation for precisely estimating cross sectional areas
of excavation or fill to minimize errors in the total earthwork cost calculation.
Aruga, Sessions and Akay (2005) [11] developed a forest road design program based on a high-resolution
Digital Elevation Model (DEM) from a light detection and ranging (LIDAR) system. After a designer had located
the intersection points on a horizontal plane, the model firstly generated the horizontal alignment and the ground
pr , and then it could precisely generate cross-sections and accurately calculate earthwork volumes using a
high-resolution DEM. A shortage of this model was that it couldn t properly optimize horizontal and vertical
alignments simultaneously. Li and Han [12] (2007) used DTM to calculate cross section area, but still completed
volume computation by 2D rule. These programs had begun to bring DTM into roadway design and volume
calculation, but actually it the completely 3D concept because they still use average-end-area or
prismoidal method to compute earthwork volume finally.
 Jian-chuan Cheng and Long-jian Jiang / Procedia - Social and Behavioral Sciences 96 (2013) 1277 – 1285 1279

Du and Teng (2007) [13] used 3D laser scanning and GPS technology to compute volume of landslide
earthwork. They employed these two advanced technology to create the contour after landslide. The original
contour of this area could be obtained from the government of the Forestry Bureau. The volume of collapse was
estimated by the difference between before and after landslide of terrain contours by the contour method with the
Simpson's rule. We consider this research as the transition form from 2D method to 3D method because it has the
3D concept but still solve the problem with 2D formula and procedure.
Bao (2011) [14] applied DTM on the earthwork calculation of land consolidation where the MapGIS DTM
analysis function was utilized to transform the elevation points into Triangulated Irregular Network (TIN) and
fixed the layout elevation by iterative calculation aimed at the balance between cut and fill. This application of
3D method could also be used in roadway design and volume calculation.
The latest report of roadway earthwork volume calculation by 3D method was conducted by Kerry, Dianne
and James [15] (2012). The researchers employed three-dimensional laser scanning to create a surface of the
original terrain based on the finite element method. Then this surface was converted to TIN file and earthwork
quantities were computed by comparing the TIN of original terrain to that of the finished project. This
methodology had already shown the 3D concept and used 3D procedure, but it was only applicable for the
computation of the final quantities of roadway. As to the design estimation, another methodology should be
developed. Still, the new 3D method was constrained by the problems such as data size, scanning accuracy and
data processing.
As reviewed above, 2D methods such as average-end-method, prismoidal formula and other models improved
based on them, are not accurate in theory but practically used in engineering. The real and accurate volume could
be got by DTM method, especially with the help of updated CAD software, GIS, GPS and laser scanning
technology (Uddin [16], 2008). More and more researchers start to transfer their focus and interest from 2D
method to 3D method (Bhatla, Choe, Fierro & Leite [17], 2012). However, 3D method is currently only applied
in the area of landslide volume computation and land consolidation. The objective of next phase is to bring 3D
method based on DTM into roadway earthwork calculation and give an accuracy comparison between 3D method
and prevailing 2D method.

3. Method and Conditions for Test Cases

DTM usually means joining the finite mass points together forming a series of surface or plane to approach the
original terrain, so its mathematical definition is the set of plane M formed by the mass points or interpolated
points Pj in area D in a specific rule (i-the number of mass points; j-the number of plane) (Tang, Liu & Lv
[18], 2005):
DTM Mi Pj | Pj (x j , y j , H j ) D, j=1, 2, , n;i 1, 2, , m (1)

When the rule is triangulation, it produces the Triangulated Irregular Network (TIN), whereas the rule is
grid it produces Grid Network (GN). Taylor, Kudowor and Fairbaim (2001) [19] proved TIN to be a more solid
and better organization method.
The original terrain surface and roadway design surface can be easily represented in TINs. Through the
operations of projection and re-triangulation between these two TINs we can get the triangle pairs as shown in
Fig.1 (in the situation of one triangle pair, plane ABC and A ' B ' C ' stand for original or design surface
respectively) (Cheng [20], 2005; Zhou & Cheng [21], 2006). Next step is to compute the volume of these
triangular prisms-when the original surface is up and design surface is down, it s excavation and contrariwise it s
fill-and sum all these volumes to obtain the final cut or fill volume.
1280 Jian-chuan Cheng and Long-jian Jiang / Procedia - Social and Behavioral Sciences 96 (2013) 1277 – 1285

Fig. 1. The cut and fill relations between two TINs

In order to analyze the accuracy difference between 2D method and 3D method, we have following cases.
The test data of plain area is an electronic geographic map with the area of 50,000 square meters and 68740
elevation points, flat, minimum elevation 36.19m, maximum elevation 60m, and average elevation 47.047m. The
test data of mountainous area is an electronic geographic map with the area of 2 million square meters in Tibet
and 211043 elevation points, hilly, minimum elevation 4110.00m, maximum elevation 5175.001m, and average
elevation 4437.492m. All the test roadways are single carriage way, 2km long, 2×3.75m wide, without central
median.
Under the different conditions of cut and fill in plain and mountainous area, we design a series of intervals by
the step length of 10m:10m,20m,30m,40m,50m,60m. In DTM method the volume result should have nothing to
-to-surface. However, this study still calculates the 3D
volume between two successive cross sections and sums them to get the final volume to compare in detail with

With the purpose of considering the influence of different longitudinal design line on the earthwork
computation result, this study comes up with the index of CRCF-Change Rate of Cut-Fill, namely the number of
intersection points between vertical alignment and original ground line in one kilometer-to evaluate the different
cut-fill condition of the roadway. Smaller value of CRCF indicates that the vertical alignment and ground profile
intersects less, so it could be the section of deep digging or high filled subgrade. Larger CRCF value on the
contrary indicates the vertical alignment is close to the ground profile but the switch between cut and fill is more
frequent.
The tests are conducted on two topographies for many times and under different roadway alignment designs.
However, this paper only shows the typical results because of the limited length of article.

4. The Accuracy Comparison Between 3D and 2D Methods

4.1 Mountainous area

Tab.1 shows the errors between two methods when CRCF=2.5 in mountainous area. The detailed errors of
other CRCF values are not listed here for reasons of space.
 Jian-chuan Cheng and Long-jian Jiang / Procedia - Social and Behavioral Sciences 96 (2013) 1277 – 1285 1281

Table 1. Errors of CRCF=2.5 in mountainous area

Station 2D (Average-end-area) 3D (DTM)

3 3
Error of Cut Error of Fill
Interval Cut (m ) Fill (m ) Cut (m3) Fill (m3)
10m 232248.79 533925.6 232168.04 533984.44 0.03% -0.01%

CRCF=2.5
234565.38 533755.97 232168.04 533984.44 1.03% -0.04%
30m 234851.84 533055.74 232168.04 533984.44 1.16% -0.17%
40m 230769.59 530812.9 232168.04 533984.44 -0.60% -0.59%
50m 217166.45 529227.56 232168.04 533984.44 -6.46% -0.89%
60m 250463 525517.78 232168.04 533984.44 7.88% -1.59%

he line charts of different CRCF values are drawn as following to easily analyze how the error of volume
relates with CRCF and interval. In Fig.2, the errors are absolute value.

(a) (b)

(c) (d)

Fig. 2. Errors of two methods with different CRCF values in mountainous area
1282 Jian-chuan Cheng and Long-jian Jiang / Procedia - Social and Behavioral Sciences 96 (2013) 1277 – 1285

From Tab.1 we can easily find that with the interval increasing from 10m to 60m, the results of DTM method
keep the same value, because DTM method calculates the volume by overlaying two surfaces. So different
intervals just section the total volume into different parts and affect the sum and accuracy of algorithm.
Fig.2 indicates that when the interval increases from 10m to 30m, the error between average-end-area method
and DTM method keeps in a low level-under 2%. However the error increases rapidly and fluctuates if the
interval increases up to 30m. This phenomenon shows that the 2D method can t get a reliable and stable result
when the interval is up to 30m. We define the range from 0 to 30m as the valid zone and range up to 30m as the
invalid zone. Comparing the charts of different CRCF values, it can be concluded that in valid zone CRCF value
has no obvious effects on the errors of two methods and the errors with different CRCF values are controlled
below 1% on the whole, no more than 2% in maximum.
The smaller the interval is the more accurate results of average-end-area method are. It can be thought that the
ground varies in linear way when the interval is small enough. Inversely with the increment of interval the ground
may not vary in linear way so the accuracy of 2D method will gradually descend as shown in the zone from 0 to
30m in the figure. When the interval increases to a critical value, two consecutive cross sections will lose the
control to the ground variation between them, which means the ground may has an abrupt change but this
exception reflect on the start and end cross sections influencing the accuracy of average-end-area method.
From the data and analysis above, it can be inferred that this critical value is about 30m and the volume will be
unexpected when the interval exceed this value.
Compared with the influence of interval on the result, CRCF value seems to have no significant effect on the
accuracy. According to the formula of average-end-area, two factors-cross section areas and interval-have effects
on the result. As long as the interval is less than critical value, 2D method is already exact enough to contain and
depict terrain changes. No matter how the vertical alignment is close to or devious to the ground profile it
influence the calculation of cross section area-the other factor that may impact on the accuracy. When it comes to
the invalid zone, CRCF is also meaningless to results. In Fig.4 the errors of different CRCF distribute randomly
because the interval which plays a more significant role in volume computation has already lost the control to
terrain change.

4.2 Plain Area

Tab.2 shows the errors between two methods when CRCF=3 in plain area. The detailed errors of other CRCF
values are not listed here for reasons of space.

Table 2. Errors of CRCF=3 in plain area

Station Average-end-area DTM

Error of Cut Error of Fill
Interval Cut m3 Fill m3 Cut m3 Fill m3
10m 26523.6 122744.57 26530.15 122667.7 -0.02% 0.06%
CRCF=3

20m 26577.41 122946.56 26530.15 122667.7 0.18% 0.23%

30m 26895.46 123208.41 26530.15 122667.7 1.38% 0.44%
40m 26520.77 123086.66 26530.15 122667.7 -0.04% 0.34%
50m 27552.69 122572.7 26530.15 122667.7 3.85% -0.08%
60m 26467.89 126466.77 26530.15 122667.7 -0.23% 3.10%
 Jian-chuan Cheng and Long-jian Jiang / Procedia - Social and Behavioral Sciences 96 (2013) 1277 – 1285 1283

The line charts of different CRCF values are drawn as following and in Fig.3 the errors are absolute value.

(a) (b)

(c) (d)

Fig.3. Errors of two methods with different CRCF values in plain area

The situation of plain area is similar to that in mountainous area: the critical value of interval lies at about
30m; however, in the valid zone (0 to 30m) the errors of 2D method are generally controlled beneath 0.5% which
is better than that in mountainous area with 1 % and the maximum error is 1.38%, also lower than that of 1.72%
in mountainous area. This comparison shows that in general the average-end-area is more accurate in plain area
than in mountainous area. The CRCF values show no significant influence on the accuracy. When CRCF values
vary from 3.0 to 7.5 in valid zone the errors are mostly kept under 1%, and 1.5% at most.

4.3 Summary

From the experiments with different CRCF values in two topographies, we can infer that the average-end-area
method has its limitations: different accuracies for different terrains, affected by many random factors and hard to
decide the reasonable interval. This study shows the critical value of interval is 30m and the CRCF value, namely
the cut-fill condition of roadway, has no significant effects on the final volume.
Contrast to the 2D method, DTM method can avoid all these problems to get the accurate results and it s
totally feasible at present by the advanced software package and computer hardware devices. This analysis
proves the DTM method to be a better way to calculate earthwork volume than average-end-area method.
1284 Jian-chuan Cheng and Long-jian Jiang / Procedia - Social and Behavioral Sciences 96 (2013) 1277 – 1285

5. Conclusions

(1) 3D method of earthwork calculation by the latest software such as Civil 3D could be easily used in practice
than ever before. 3D method possesses a higher accuracy than 2D method without consideration the error of the
raw terrain data.
(2) Average-end-area method performs a higher accuracy in flat area than in mountainous area. There exists a
critical station interval 30m which means 2D method could keep a relevant high accuracy below this value.
What s interesting is the CRCF value, an index firstly proposed to represent the cut-fill variance frequency
associated with roadway terrain and alignment design, has no significant impacts on the accuracy. In other words
2D method with appropriate interval is suitable for different kinds of design plans. These all reconfirm why 2D
method could be used worldwide although it has limitations.

References

[1] Hintz, C., & Vonderohe, A.P. (2011). Comparison of Earthwork Computation Methods. Proceeding of TRB 2011 Annual Meeting,
Washington DC,.
[2] Easa, S. (1988). Selection of Roadway Grades that Minimize Earthwork Cost Using Linear Programming. Transportation Research,
22(2), 121-136.
[3] Moreb, A.A. (1996). Linear Programming Model for Finding Optimal Roadway Grades that Minimize Earthwork Cost. European
Journal of Operational Research, 93,148-154.
[4] Goktepe, A.B., & Lav, A.H. (2003). Method for Balancing Cut-Fill and Minimizing the Amount of Earthwork in the Geometric
Design of Highway. Journal of Transportation Engineering, 129(5), 564-571.
[5] Easa, S. (1989).
(October, 1987, Vol. 113, No. 3). Journal of Surveying Engineering, 115(3), 356-359.
[6] Easa, S. (1992). Discussion of Cut and Fill Calculations by Modified Average-End-Area Method by James W. Epps and Marion W.
Corey (September-October, 1990, Vol. 116, No. 5). Journal of Transportation Engineering, 118(4), 600-601.
[7] Anderson, J. M., Hikhail, E. M., & Woolnough, D. F. (1985). Introduction to surveying, New York: McGraw-Hill.
[8] Easa, S. (1992). Estimating Earthwork Volumes of Curved Roadways-Mathematical Model. Journal of Transportation Engineering,
118(6), 834-849.
[9] Easa, S. (2003). Estimating Earthwork Volumes of Curved Roadways-Simulation Model. Journal of Surveying Engineering, 129(1),
19-27.
[10] Kim, E., & Schonfeld, P. (2001). Estimating Highway Earthwork Cross Sections Using Vector and Parametric Representation.
Proceeding of TRB 2001 Annual Meeting, Washington DC,.
[11] Aruga, K., Sessions, J., & Akay, A.E. (2005). Application of an Airborne Laser Scanner to Forest Road Design with Accurate
Earthwork Volumes. Journal of Forest Research, 10(2), 113-123.
[12] Li, Z.Q. & Han, F. (2007). Fast Calculation Method for Cubic Metering of Earth and Stone of Road Line Based on GIS. Journal of
Lanzhou University of Technology, 33(5), 128-131.(In Chinese)
[13] Du, J.C., & Teng, H.C. (2007). 3D Laser Scanning and GPS Technology for Landslide Earthwork Volume Estimation. Automation in
Construction, 16, 657-663.
[14] Bao, Y. (2011). Applications of the MapGIS-based Digital Topographic Model on Earthwork Calculation on Land Consolidation.
Proceeding of International Conference on Computer Distributed Control and Intelligent Environmental Monitoring, China.
[15] Kerry, T.S., Dianne, K.S., & James, P.P. (2012). Road Construction Earthwork Volume Calculation Using Three-Dimensional Laser
Scanning. Journal of Surveying Engineering, 138(2), 96-99.
[16] Uddin, W. (2008). Airborne Laser Terrain Mapping for Expediting Highway Projects-Evaluation of Accuracy and Cost. Journal of
Construction Engineering and Management, 134(6), 411-420.
[17] Bhatla, A., Choe, S.Y., Fierro, O., & Leite, F. (2012). Evaluation of Accuracy of As-built 3D Modeling from Photos Taken by
Handheld Digital Cameras. Automation in Construction, 28, 116-127.
[18] Tang, G.A, Liu, X.J., & Lv, G.N. (2005). Digital Elevation Model and Geoanalysis: Priciples and Methods. Beijing: Science Press.
(In Chinese).
[19] Taylor, G., Kudowor, A.Y.T., & Fairbairn, D. (2001). Empirical Method of Error Analysis in Volume Estimation-Monte Carlo
Simulation under Kriging and Triangulation. Journal of Surveying Engineering, 127(2), 59-77.
[20] Cheng, J.C. (2005). 3-Dimensional Calculation of Roadway Earthwork Volume. Journal of Southeast University, 21(1), 88-91. (In
Chinese)
 Jian-chuan Cheng and Long-jian Jiang / Procedia - Social and Behavioral Sciences 96 (2013) 1277 – 1285 1285

[21] Zhou, C., & Cheng, J.C. (2006). Earthwork Volume Calculation based on Digital Multi-layer Terrain Model. Proceeding of the 2006
Annual Conference of China Highway & Transportation Society, China. (In Chinese)