Professional Documents
Culture Documents
Waste Management
journal homepage: www.elsevier.com/locate/wasman
a r t i c l e i n f o a b s t r a c t
Article history: In this study, the potential re-use of waste foundry sand in high-strength concrete production was inves-
Received 28 June 2009 tigated. The natural fine sand is replaced with waste foundry sand (0%, 5%, 10%, and 15%). The findings
Accepted 9 February 2010 from a series of test program has shown reduction in compressive and tensile strengths, and the elasticity
Available online 9 March 2010
modulus which is directly related to waste foundry inclusion in concrete. Nevertheless the concrete with
10% waste foundry sand exhibits almost similar results to that of the control one. The slump and the
workability of the fresh concrete decreases with the increase of the waste foundry sand ratio. Although
the freezing and thawing significantly reduces the mechanical and physical properties of the concrete.
The obtained results satisfies the acceptable limits set by the American Concrete Institute (ACI).
Ó 2010 Elsevier Ltd. All rights reserved.
0956-053X/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.wasman.2010.02.018
1706 Y. Guney et al. / Waste Management 30 (2010) 1705–1713
Several researchers have studied foundry sand in recent years, Aggregate Grain size Specific Water
some of whose work has been listed in the following: use of foun- (mm) gravity content (%)
dry sand in highways and sea barriers (Kirk, 1998; Mast and Fox, Coarse Crushed stone-1 8–16 2.64 0.4
1998); hydraulic barriers (Abichou et al., 1998, 2000); asphalt con- Crushed stone-2 4–8 2.63 0.3
Fine Natural sand 0–4 2.60 2.0
crete (Bakis et al., 2006); investigating the permeability with slurry
material, usage opportunity as waste material in civil engineering
and recycling (TRB, 1994; Reddi et al., 1995); quality of leakage
water in the foundry sand landfill areas and its properties (Ham aggregates (American Foundrymen’s Society, 2004). With respect
et al., 1986; Institute of British Foundrymen, 1996a,b); recycling to these points, two different sizes of crushed stone (as a coarse
and removal off waste foundry (Senske, 1970). In addition, Khatib aggregate), natural sand, and waste foundry sand (as a fine aggre-
and Ellis (2001), Naik et al. (2003, 2004), Fiore and Zanetti (2007), gate) are arranged for the preparation of the mixture. The maxi-
Siddique et al. (2007, 2009), and Siddique and Noumowe (2008) mum grain size in the mixtures is 16 mm. The grain size, specific
have reported the use of waste foundry sand in concrete and con- gravity, and the water content of the aggregates have been pro-
crete-related products. Guney et al. (2006) investigated the re-use vided in Table 1.
of foundry sand modified mixtures for sub-based layers of high-
ways. Naga and El-maghraby (2003), Raupp-Pereiraa et al.
3.1.2. Waste foundry sand and cement
(2006), Quaranta et al. (2004), and Seung-Whee and Woo-Keun
The waste foundry green sand is obtained from the Toprak Cast-
(2006) have reported the use of waste foundry sand in ceramic
ing Industry Bozuyuk Facilities in Eskisehir, Turkey. This sand is
materials and tile making process. Korac et al. (2006) investigated
used as foundry sand in steel and metal molding facilities for the
the used-foundry sand in making of plaster mixture. El Haggar and
production of metal-steel parts at high temperatures (1500 °C)
El Hatow (2009) investigated the use of rejected and un-recyclable
for about 8–10 times until loosening its molding properties. The
plastics with used-foundry sand for manhole covers and its base.
Scanning Electron Microscope, SEM, photography of foundry sand
Gao and Drummond (1999), Ferraris et al. (2001), and Colombo
is shown in Fig. 1. The waste foundry sand used in the current
et al. (2003) have reported the use of waste foundry sand for the
study is generally sub-angular to round in shape. The grain size
inertization and re-use of waste materials by vitrification. In addi-
tion, Raupp-Pereiraa et al. (2004) investigated the refractory mor-
tars made of waste foundry sand.
3. Experimental studies
3.1. Materials
3.1.1. Aggregate
Aggregate properties strongly influence freshly-mixed and
hardened properties the concrete. On the other hand, specifications
governing the selection and use of aggregates in concrete mixtures Fig. 1. (a) The SEM photography of a single foundry sand piece. (b) The SEM
generally relate to the particle-size distribution or gradation of the photography of foundry sands.
Y. Guney et al. / Waste Management 30 (2010) 1705–1713 1707
distribution of waste foundry sand is uniform, with 100% of the is due to the fact that the ACI 318-99 standard requires a maxi-
material under 1 mm, 10% of foundry sand greater than 0.5 mm mum water-to-cement ratio (w/c) of 0.45 by mass for concrete to
and 5% smaller than 0.125 mm; it is black in color (the sea-coal or- be exposed to freezing–thawing in moist conditions. The aggregate
ganic binder has been used in the foundry sand) and has a specific ratios used in the mixtures are shown in Fig. 2.
gravity of 2.45. The results of the chemical analyses, X-ray fluores-
cence (XRF) and atomic absorption spectrometry (AAS) of the 3.2.2. Experiments
waste foundry sand are shown in Table 2. The ASTM Type I, CEM The workability of the fresh concrete was measured by the
cement is used as a binding material. The physical properties and slump test according to the ASTM C143 standard. Standard test
chemical composition of the typical foundry sand, waste foundry method for splitting tensile strength of concrete specimens accord-
sand, and the cement have also given in Table 2. ing to the ASTM C496 was conducted to determine such strength;
the ASTM C39 standard test method was used to test the compres-
3.2. Laboratory studies sive strength; the ASTM C597 test method for the determination of
the ultrasonic pulse velocity of propagation of compression waves
3.2.1. Mixtures and properties in concrete; and the ASTM C642 test method that sets specific
The ASTM C33 allows a maximum of 5% fine aggregate particles gravity, water absorption, and void ratio in hardened concrete,
to pass through sieve No. 200. These particles include clay and was conducted at the age of 28 and 56 days. After 120 days of cur-
other dusts. High dust content can interfere with the bonding of ing, the specimens were frozen in a temperature chamber at
the cement onto the surface of the aggregate, and can also increase 24 ± 1 °C for 24 h, and then thawed according to the ASTM
water demand. These factors reduce the durability of hardened C666. The unconfined compressive strength and the dynamic elas-
concrete. The particle-size distribution of foundry sand was pre- ticity modulus were, then, measured before and after 80 freeze–
cisely arranged according to these regulations. The waste foundry thaw cycles.
sand passing the sieve No. 200 is 4% which satisfies the ASTM
C33 requirement. Apart from this, mixtures were set up in such a 4. Results and discussion
way that the particle-size distributions are in between the A16
and B16 reference curves according to the Turkish Standard 4.1. Slump test
TS706 (1980) (Fig. 2). The fineness modulus of the used-foundry
sand is 2.53. The slump test results of the fresh concrete to reveal the effects
The waste foundry sand is replaced with 5%, 10%, and 15% of the of the waste foundry sand on the consistency are given in Fig. 3. It
fine natural sand by mass. The amount of cement used is 500 kg/ is observed that the waste foundry sand decreases the fluidity and
m3, and w/c ratio of the mixtures is proportioned to be 0.45. This the slump value of the fresh concrete. This is probably due to the
presence of clay-type fine materials in the waste foundry sand,
which are effective in decreasing the fluidity of the fresh concrete.
Table 2
Physical properties and chemical composition of the waste foundry sand, cement and
typical foundry sand. 4.2. Compressive strength and modulus of elasticity
Waste Cement (ASTM Typical
foundry sand Type I, CEM) foundry sand* The unconfined compression test results for the various
Physico-chemical properties, value
amounts of waste foundry sand as partial replacement of fine
Gs 2.45 3.1 2.39–2.55 aggregate at 28 and 56 days have been plotted in Fig. 4. The mod-
Cu 5.5 – – ulus of elasticity is also calculated according to the equation sug-
Fines content 24 – – gested by the ACI 318 at the age of 28 and 56 days, as follows:
(<74 lm) (%)
Active clay content 5 – – E ¼ 0:043 W 3=2 r1=2 ð1Þ
(<2 lm) (%)
Moisture content (%) 3.25 – 0.1–10.1 where E is the modulus of elasticity in MPa, W is density in kg/m3,
pH 9.1 – –
Organic content (%) 4.3 – –
and r is the unconfined compressive strength in MPa. The calcu-
lated moduli of elasticity have also been provided in Fig. 4. It can
Chemical composition, value (%)
Silica (as SiO2) 98 20.35 87.91
be seen, from Fig. 4, that the compressive strength increases with
Aluminum (as Al2O3) 0.8 5.19 4.70 age for all the tested mixtures. It is also observed that the concrete
Iron (as Fe2O3) 0.25 3.35 0.94 with 10% waste foundry sand replacement exhibits highest com-
Potassium (as K2O) 0.04 – 0.25 pressive strength at the age of 56 days. The compressive strength
Calcium (as CaO) 0.035 64.56 0.14
decreases with an increase in the amount of foundry sand. The con-
Magnesium (as MgO) 0.023 1.52 0.30
SO3 0.01 3.1 0.09 crete with 10% waste foundry replacement may indicate the opti-
Na2O 0.04 0.1 0.19 mum re-allocation amount of waste foundry sand in this study.
TiO2 – 1.1 0.15 The concrete with 10% waste foundry sand shows almost the same
Mn2O3 0.01 – 0.02 strength as that in the control concrete mixture whereas all the
LOI – 1.4 5.15
other waste foundry-added mixtures exhibit lower values than that
Nickel 0.004 – –
Chromium 0.003 – – of the control. This may indicate that the particle-size distribution
Lead 0.003 – – of the mixture with 10% waste foundry sand has sufficient adher-
Zinc 0.003 – – ence than the other mixtures with waste foundry sand. On the other
Copper 0.002 – –
hand, it is obvious from Eq. (1) that the static modulus of elasticity
Cadmium 0.001 – –
Sulphur – – – is a function of the compressive strength of the concrete. If the com-
Trace elements 0.836 – – pressive strength of the concrete increases, the static modulus of
elasticity will also increase, and vice versa. This fact can be clearly
Note: Gs: specific gravity, Cu: coefficient of uniformity (=D60/D10). Fines content
and grain size diameters (for Cu calculations) are based on ASTM D422. Active clay
observed in the Fig. 4.
content was determined following the procedures in ASTM C837. Reddi et al. (1995) reported that the compressive strength of
*
American Foundrymen’s Society (2004). stabilized waste foundry sands decreases with the increase in
1708 Y. Guney et al. / Waste Management 30 (2010) 1705–1713
100
Under Sieve, %
80
60
40
20
0
0 0.25 0.5 1 2 4 8 16
A16 0 3 7.5 12 21 36 60 100
B16 0 8 20 32 42 56 76 100
K0 0 7 14 23 34 48 69 100
K5 0 7 14 23 34 48 69 100
K10 0 7 14 23 33 48 69 100
K15 0 7 14 23 34 48 69 100
Sieve Size, mm
Fig. 2. The particle-size distribution of the mixtures (A16, B16 Turkish standard; K0, K5, K10, K15 waste foundry mixes).
17
15
13
Slump, cm
11
5
0 5 10 15
Waste Foundry, %
Fig. 3. The effect of the waste foundry sand on the slump behavior.
waste foundry sand content in the mixtures. This while, the clay- strength increased marginally with the increase in foundry sand. At
bonded foundry sand reduces the strength of the stabilized mixes the age of 28 days, the compressive strength increased by 4.2%,
more than the chemically-bonded foundry sands. 5.2%, and 9.8% when compared to control mix without foundry
Siddique et al. (2007) have conducted a study concerning the ef- sand, whereas the increase was 1.0%, 5.18%, and 14.3% at the age
fect of waste foundry sand as partial replacement of fine aggregate of 56 days.
on the medium compressive strength of the concrete. The control Khatib and Ellis (2001) carried out a research on the compres-
mixture without waste foundry sand was proportioned to com- sive strength of concrete prepared with three types of foundry sand
pressive strength of about 30 MPa at a 28-day period. They con- as a partial replacement for fine aggregate. Three types of sand
cluded that the compressive strength of concrete mixes with were used in foundries: the white fine sand without the addition
10%, 20%, and 30% fine aggregate replacement with foundry sand of clay and coal, the blended foundry sand (before casting), and
was higher than the control mix at all ages. Also, the compressive the waste foundry sand. The standard fine sand was partially re-
Y. Guney et al. / Waste Management 30 (2010) 1705–1713 1709
65
64
40
63
62
Compressive Strength, MPa
60
59
38
58
57
56 37
55
54
36
53
52
51 35
0 5 10 15
Waste Foundry Sand, %
Fig. 4. Effect of waste foundry sand on compressive strength and modulus of elasticity.
placed with these types of foundry sand (25%, 50%, 75%, and 100%). sult has to be verified with additional studies concerning added
The increase in the replacement level of standard sand with the replacement of waste foundry sand percentage ratios around
foundry sand decreased the strength of the concrete linearly. The 10%. The mixtures containing more than 10% used-foundry sand
concrete with white sand replacement showed similar strength had compressive strength lower than the control mixture; hence,
to those containing waste foundry sand. The high replacement of these mixtures are not appropriate for high-quality structural
the blended foundry sand caused a reduction in strength as com- concrete.
pared with concrete incorporating white sand or waste foundry
sand. 4.3. Splitting tensile strength
Naik et al. (2004) studied the influence of waste foundry sand as
well as fly ash and bottom ash on the compressive strength of The tensile strength characteristics have considerable impor-
bricks, blocks, and paving stones. The replacement rates by mass tance in any type of high-strength concrete design procedure.
for sand with waste foundry sand were 25% and 35%. They con- The results show that the waste-foundry-sand-included mixtures
cluded that the replacement of that part of the sand with waste develop substantial tensile strength that shows a close relationship
foundry sand in mixtures had caused a reduction in the strength. with the content of such sand. The splitting tension test results of
It was also determined that 25% of the fine sand in blocks could the specimens, with and without the foundry sand, are given in
be replaced with waste foundry sand in cold regions, while and Fig. 5. The similar results – like those in the compression test –
up to 35% of the sand in bricks and blocks could be replaced with are obtained from the splitting tension test. The splitting tensile
waste foundry sand for use where frost action is not a concern. strength values of 5% and 15% waste-foundry-sand-replaced-spec-
The results stated by Khatib and Ellis (2001) and Naik et al. imens are lower than that of the control one; the specimens re-
(2004) are similar to the present study except the case with 10% placed with 10% waste foundry sand have slightly higher values
of waste foundry sand replacement which increases the compres- than the control mix (without foundry sand). The results of the
sive strength and modulus elasticity. Unlike the current study, Sid- splitting tension test can be considered as acceptable values when
dique et al. (2009) stated that the compressive strength increased compared to that of the ACI 318 relation as in,
with the increase in waste foundry sand for all ages for all mixes.
They stated that the increase in compressive strength with the fsp ¼ 0:47ðfc0 Þ0:56 for 5 MPa < fc0 < 120 MPa ð2Þ
inclusion of used-foundry sand could probably be due to the fact
that used-foundry sand was finer than regular sand which resulted were fsp is the splitting tensile strength in MPa, and fc0 is the com-
in the denser concrete matrix, and also due the silica content pres- pressive strength in MPa.
ent in the used-foundry sand. In addition, Siddique and Noumowe Siddique et al. (2007) have studied the effect of waste foundry
(2008) stated that clay-bonded foundry sand reduces the strength sand as partial replacement of fine aggregate on the splitting ten-
of stabilized mixes more than the resin-bonded sands owing to the sile strength and the elasticity modulus of the concrete. The pro-
absorption of water by the clay content. portions of the replaced fine aggregate were 10%, 20%, and 30%.
In this study, the 10% amount of replacement of waste foundry Their results indicated that the splitting tensile strength increased
sand initiates a different behavior than some of the previous stud- with an increase in the foundry sand. The modulus of elasticity for
ies stated in the literature. This may imply optimum re-allocation foundry-sand-concretes with 10%, 20%, and 30% fine sand replace-
of the waste foundry sand for this investigation. This contrary re- ment was higher than the control mix at all ages. They also ob-
1710 Y. Guney et al. / Waste Management 30 (2010) 1705–1713
3.8
3.6
3.4
3.2
3.0
0 5 10 15
Waste Foundry Sand, %
Fig. 5. Effect of waste foundry sand content on splitting tensile strength and ACI 318 calculations.
served that the modulus of elasticity of in the all mixes continued has decreased for the specimens having waste foundry sand of
to increase with age. 10% and 15%. This may be explained as the consumption of waste
The use of waste foundry sand in asphalt concrete with regard foundry sand decreases the voids in the concrete. Therefore, water
to tensile strength has been investigated by Bakis et al. (2006). absorption values have a tendency to decrease in the specimens
The asphalt concrete mixtures were prepared with 0%, 4%, 7%, with greater waste foundry sand than 5%. It is also observed that
10%, 14%, 17%, and 20% replacement of fine sand with waste foun- the water absorption ratios of the specimens supplied with 10%
dry sand. Tests were performed for splitting tensile strength as per waste foundry sand and the specimens without waste foundry
the AASTHO T283 standard. The strength of the asphalt concrete sand are close to each other. The specimens supplied with waste
mixtures showed a nearly linear decrease as the percentage of foundry sand a have lower absorption capacity than the control
waste foundry sand increased (1.39 MPa with 0% waste foundry ones. The changes in the void ratios of concrete samples with
sand and 0.94 MPa with 20% waste foundry sand). and without waste foundry sand are also given in Fig. 6. The re-
In the current investigation, the tensile strength of all the mix- sults of the void ratio of the samples with and without waste foun-
tures decreased with the addition of waste foundry sand (except dry sand are similar to the results water absorption test. The
for that with partial substitution of 10% waste foundry sand). This replacement of waste foundry sand with fine sand in ratios greater
means that the results here are similar to those of the Bakis et al.’s than 5% decreases the void ratios for all ages. From their studies,
(2006). Although, the results of Siddique et al. (2007) have shown Siddique and Noumowe (2008) and Siddique et al. (2009) have
that the splitting tensile strength increases to remarkable level concluded that waste foundry sand exhibits higher water absorp-
when compared to ordinary mix without foundry sand, the tion level and higher void percentage compared to regular con-
replacement of waste foundry sand exhibits an adverse effect in crete sand. It is worth noting that the findings of this study are
this study other than a particular substitution amount of 10%. not similar to the predictions of Siddique and Noumowe (2008)
The test results of the mixture containing this amount of substitu- and Siddique et al. (2009).
tion are higher than the ACI 318 estimated tensile strength values
for all ages (see Fig. 5). 4.5. Freezing–thawing resistance and dynamic elasticity modulus
9.0
3.4
8.5
3.2
8.0
Water Absorption, %
3.0
7.5
Ratio, %
2.8
7.0
Void
2.6
6.5
2.4
6.0
2.2 5.5
2.0 5.0
0 5 10 15
Waste Foundry Sand, %
Fig. 6. Effect of waste foundry sand content on water absorption and void ratio.
70
51
65
50
60
49
55
48
50
47
45
40 46
0 5 10 15
Waste Foundry Sand, %
V 2 n ð1 þ lÞ ð1 2lÞ freezing and thawing cycles. The same trend has been noted in all
Ed ¼ ð3Þ the mixes; also, the concrete samples with 10% foundry sand mix-
1l
ture are slightly influenced by freezing and thawing cycles with re-
where Ed is the dynamic modulus of elasticity (MPa), V is the ultra- spect to the other waste foundry sand replacement ratios. On the
sonic pulse velocity (m/s), n is the unit weight of the specimen (kg/ other hand, the ACI 318-99 code sets a minimum specified com-
m3), and l is the Poisson’s ratio (0.15). pressive strength of 31.0 MPa for concrete exposed to freezing
The obtained results for the dynamic modulus of elasticity, Ed, and thawing in a moist condition. All the mixes have sufficiently
are plotted in Fig. 7. It can be observed that both the compressive high compressive strengths to comply with the specified value of
strength and the dynamic modulus of elasticity decrease with the the ACI 318-99.
1712 Y. Guney et al. / Waste Management 30 (2010) 1705–1713
The effect of freezing and thawing on bricks, paving stones and worthwhile to note that the freezing–thawing influences on
blocks was studied by Naik et al. (2004). Their tests were performed both of the used-foundry-sand-added and control specimens
according to the ASTM C 1262. Based on whose results, they re- were about the same. It is also important to note that the labo-
ported that partial substitution of conventional sand with waste ratory-simulated freezing and thawing conditions (ASTM C 666)
foundry sand caused an almost disregardable effect on strength, are harsher than the normally encountered winter conditions in
density, and absorption; blocks with 25% sand replacement showed most regions in Turkey, and most of the mixtures are likely to
a large reduction in the freezing and thawing period compared to exhibit a better performance in the field.
the fly ash. Also, blocks with 35% replacement of sand with either The transportation costs are generally the highest for the re-use
bottom ash or waste foundry sand showed a very sharp reduction of waste foundry sand. The most economically-sustainable
in the above-mentioned period compared with the fly ash. options seems to be to set up construction projects at locations
Although the strength of blocks containing waste foundry sand closest to waste foundry sand stock areas of both appropriate
was considerably higher than of those containing bottom ash, the quality and sufficient quantity. In this way, the use of such sand
freezing and thawing behaviors of the two groups of blocks were in high-strength concrete will be beneficial.
about the same. In the current study, and after 80 standardized The results showed that foundry sand can be successfully used
freeze–thaw cycles, it was found that the concrete with 10% content in high-strength concrete applications if the particle-size distri-
of waste foundry sand showed good resistance to frost period. In bution is arranged carefully, thus providing similar properties to
fact, a decrease in the compressive strength and the dynamic mod- that of the high-strength concrete containing standard fine sand.
ulus of elasticity did not exceed 10% and 7%, respectively. The other It can be inferred from these results that the use of waste foun-
mixtures, however, showed higher reduction in the compressive dry sand benefits the environment as well since such stocks can
strength than this mix – a decrease of 11% and 18% for 5% and be put to profitable use and the transportation cost and require-
15% replacement of foundry sand, respectively. ments are reduced considerably.
Acknowledgements
5. Conclusions
This research has been supported by the Anadolu University
According to the results of this study, the following conclusions
Research Fund under Project Number 031010.
can be drawn;
Naik, T.R., Kraus, R.N., Chun, Y.M., Ramme, W.B., Singh, S.S., 2003. Properties of field Seung-Whee, R., Woo-Keun, L., 2006. Characteristics of spent foundry sand–loess
manufactured cast-concrete products utilizing recycled materials. Journals of mixture as ceramic support materials. Materials Science Forum 510-11, 378–
Materials in Civil Engineering 15 (4), 400–407. 381.
Naik, T.R., Kraus, R.N., Chun, Y.M., Ramme, W.B., Siddique, R., 2004. Precast concrete Senske, M.L., 1970. Waste Molding Sand Management Practices, Office of Solid
products using industrial by-products. ACI Materials Journal 101 (3), 199–206. Waste, Environmental Protection Agency Old Reports TO-5/0, p. 15.
Raupp-Pereiraa, F., Nunes, A.F., Segadães, A.M., Labrincha, J.A., 2004. Refractory Siddique, R., Gupta, R., Kaur, I., 2007. Effect of spent foundry sand as partial
mortars made of different wastes and natural sub-products. Key Engineering replacement of fine aggregate on the properties of concrete. In: 22nd
Materials 1743–1747, 264–268. International Conference on Solid Waste Technology and Management
Raupp-Pereiraa, F., Hotzab, D., Segadãesa, A.M., Labrincha, J.A., 2006. Ceramic Widener University, Philadelphia, USA.
formulations prepared with industrial wastes and natural sub-products. Siddique, R., Noumowe, A., 2008. Utilization of spent foundry sand in controlled
Ceramics International 32, 173–179. low-strength materials and concrete. Resources, Conservation and Recycling 53,
Quaranta, N., Caligaris, M., López, H., Unsen, M., Pasquini, J., Lalla, N., Boccaccini, 27–35.
A.R., 2004. Recycling of foundry sand residuals as aggregates in ceramic Siddique, R., Schutter, G., Noumowe, A., 2009. Effect of used-foundry sand on the
formulations for construction materials. Key Engineering Materials 264–268, mechanical properties of concrete. Construction Building Materials 23, 976–980.
1743–1746. TS706 Turkish code 706, 1980. Aggregates for Concrete. TSE Ankara, Turkey (in
Reddi, N.L., Rieck, P.G., Schwab, A.P., Chou, S.T., Fan, L.T., 1995. Stabilization of Turkish).
phenolics in foundry sand using cementitious materials. Journal of Hazardous TRB, 1994. Aggregates: Waste and Recycled Materials; Soils, Geology and
Materials 45, 89–106. Foundations; Materials and Construction, National Academy of Science
Transportation Research Board, TRB/TRR-1437, p. 73.