TCP

“Thin Concrete Pavements”

Juan Pablo CovarrubiasT Civil Engineer, PhD, MsC Juan Pablo Covarrubias V Civil Engineer December 2007

Abstract

It is important to understand the behavior of concrete pavements. Measurements

made in Chile reveal that slabs are always curled with the borders in the air and never

warped, the same thing happens in Guatemala. This curling is mostly influenced by

thermal and hydraulic gradients during construction. This condition of the slab is opposed

to the one assumed for the normal design procedures used until now. Because of this

condition, the maximum tensile stresses are on the top of the slab.

The size of a slab in JPCP concrete pavements has an important effect on the

stresses in the concrete. These stresses are the cause of fatigue life and affect the

performance of the pavement. A new way of designing concrete pavements is to

minimize the stresses by considering the way trucks load the slab. Designing the slabs

sizes so that the trucks never apply more than one set of wheels on any slab, reduces the

stresses imposed by them. The smaller size also reduces curling.

Improvement of these factors can reduce stresses and allow the design of thinner

slabs for the normal performance of concrete pavements. A slab size of half a lane wide

and 1.4 m long, which is smaller than the distance between axels of the tandem rear axels

of trucks, can reduce the thickness by more than 10 cm (4 inches) for the same stresses

within the slab as for 4,5 m long slabs, therefore the same performance in time.

For thick slabs the main stresses are on the top of the slab, while for thin slabs (14

cm or less) the main stresses are on the bottom. The effect of this is that the effect of the

stiffness of the soil layers is different for thick and thin slabs. For thin slabs the base has

to be designed, as it has an effect in reducing the stresses in the slab. A summary on how

concrete slabs deform, a discussion on the cause of the stresses and an analysis made with

finite elements with ISLAB2000 for new concrete pavements are discussed. The analysis

considers a concrete pavement over a granular sub-base and a given truck configuration

and loading. The results show tensile stresses in the concrete for different relations

between length and thickness of the slabs.

In the same way, a reduction in size with a given thickness will allow for higher

axel loads for any given stress. This higher load capacity could mean an increase of axel

loads allowed and a reduction in number of trucks, contamination and transportation

costs.

As a conclusion, it can be said: TCP design has lower stresses for a given

thickness allowing higher axel loads or slabs can be thinner for a given stress with actual

axel loads.

TCP technology (Thin Concrete Pavements), the methodology for the design and construction of improved concrete pavement slabs and other rights related to this technology (software, know-how, industrial secrets, trademarks, manuals, instructions, etc.), are exclusively owned by Comercial TCPavements Ltda. and subject to legal protection as recognized in local regulations and International Intellectual and Industrial Property Treaties, particularly by patent applications Nos. 2684-05 in Chile, international application PCT/EP2006/064732, application 20070094990 in the United States. ©TCPavements 2005-2007, all rights reserved.

DEVELOPMENT:

Analyzing the performance of concrete pavements and its relation to curling, there are

some thoughts that can be discussed. In Chile there was a very bad experience of

unbonded slabs over cement treated bases. A polyethylene sheet was placed between the

slab and the CTB. The cracking of these pavements started in about eight years, while in

pavements of the same contract over granular bases, with the same polyethylene under

the concrete, the cracks started after fifteen years. It was also seen that shorter slabs

performed better than longer slabs. This performance shows the effect of bonding,

rigidity of the base and length of slabs. The following thinking tries to explain this

performance and optimize concrete pavement design.

A) Effect of the Rigidity of the Base.-

The pavement slabs are supported by the base. When the slab curls with the edges

upward, if the base is stiff, the slab will not sink on it and the central area of support will

be small and the cantilever long (Figs. 1 and 2). With the loads at the edges, this will

produce high tensile stresses on the surface of the slab and top down cracks. If the base

has the appropriate stiffness, the slab will sink on it leaving a shorter cantilever and lower

stresses with the same loading. For this case, the ideal support rigidity is with a stiffness

of CBR 20 to 50%.

If the base is too soft with the load at the center, tensile stresses at the bottom of the slab

and bottom up cracks will be produced. This is explained because the slab will be wholly

supported and the stresses will be induced by the deformation of the slab over a

deformable support (Fig 3). This same effect is induced if the slabs are warped

downward. This is the original thought on calculating the stresses with the design

methods, before the curling up phenomena was known.

Fig. 1 .-Measured curling on an industrial floor slab 150 mm thick, 4 meters long. The

slab is supported on the central circle, the edges are in cantilever. The corners are four

times more deformed than the center of the edges. (Holland 2002)

Debonded, with polyethylene between CTB and concrete slab

Fig 2.- Effect of stiffness of the base on cantilever length on debonded concrete slabs.

Fig. 3.- Effect of base stiffness on amount of cracking in slabs. A medium stiffness is

better than very stiff or very soft. The optimum is between CBR 30 to 50% (Armanghani

1993).

This suggests that the optimal material to use as base material would be with CBR

between 20 and 50% when long slabs are curled upwards. In Chile, the most durable

concrete pavements (more than 70 years on a high traffic road) were built over bases with

CBR 30%.

The needed stiffness of the base could be different if the slabs are shorter, thinner and

flatter, with the tensile stresses at the bottom with bottom up crack possibility. If the slabs

are thinner and more flexible, the stresses at the bottom can be high.

B) Effect of Slab Length on the Cantilever Length and Tensile Stresses.

The curling is produced by a force on the surface of the slab. This force is due to drying

and thermal differential shrinkage on the surface of the concrete. This force, being at the

surface and not at the neutral axis of the slab, produces the curling. The magnitude of this

force is dependant on the length of the surface, therefore is smaller on shorter slabs,

which means that shorter slabs have reduced curling.

Also, if the slab curls upward leaving a cantilever of a given fraction of its length, then a

shorter slab will have a shorter cantilever (Fig 4)

These two effects mean that shorter slabs will have reduced curling and shorter

cantilevers, so, reduced tensile stresses on the top than longer slabs.

Fig 4.- Shorter slabs have shorter cantilevers than longer slabs, and therefore

smaller tensile stresses on the top.

C) Effect of Hydraulic and Thermal Differentials in Curling and Cantilever Length

The drying shrinkage curling is due to the hydraulic difference between the top and the

bottom of the slab. The slab is always wet at the bottom, as the humidity of the earth

condenses under the pavement, and it is most of the time dry on the surface. This

humidity gradient produces an upward curling (Fig 5). The residual upward curling for

the slab with cero temperature gradient was measured in Chile on real pavements, and

was equivalent to a thermal gradient of 17.5 °C with the top surface colder. The

maximum positive gradient measured at midday, when the slab is hottest at the surface,

was 19 °C. This means that the slab got flat on the ground just for a short period of the

day. In most of the time presented an upward curling, being maximum at night, when the

built in and the negative temperature gradient, with the top colder, are added. This gives

the maximum upward curling of a slab, and normally occurs at early hours in the

morning, before the sun rises.

Construction is important to reduce inbuilt hydraulic curling. A good curing to prevent

surface water loss when the concrete is not stiff enough will reduce curling. Allowing

some drying of the concrete from the bottom surface of the slab, by not using

impermeable materials under the slab or not saturating the base before placing the

concrete, also reduces humidity curling. Care should be taken on temperature of the base

when placing the concrete. Maybe some watering should be done to reduce the

temperature of the base.

The main thermal shrinkage is produced during construction. When the concrete is placed

during the hot hours of the day, the concrete on the surface of the slab will be hotter and

will harden with a longer surface because of its higher temperature than the bottom

surface. It will also harden first. When the temperature comes down to normal working

temperature, the top of the slab will reduce its length more than the bottom part, and

induce a superficial force that produces the upward curling. Placing the concrete in the

afternoon and evening, will reduce high surface temperatures when the concrete sets, and

will reduce curling due to thermal differentials.

These forces induced by drying and temperature shrinkage of the surface are dependant

of the slab length. For longer slabs, the curling force will be bigger, and so the curling

and the cantilever.

1< a ≤ 2

Fig 5.- Schematic forces, including curling lifting forces, in a concrete slab.

450 cm x350 cm Thickness= 30 cm Delta T°=-14 C° Linear Curling= 0,165 Top Tensile stress= 7,20 kg/cm2

450 cm x350 cm Thickness= 25 cm Delta T°=-14 C° Linear Curling= 0,1915 Top Tensile stress= 9,24 kg/cm2

450 cm x350 cm Thickness= 18 cm Delta T°=-14 C° Linear Curling = 0, 2416 cm Top Tensile stress= 14,02 kg/cm2

350 cm x350 cm Thickness= 18 cm Delta T°=-14 C° Linear Curling= 0,2015 Top Tensile stress= 9,32 kg/cm2

240 cm x175 cm Thickness= 18 cm Delta T°=-14 C° Linear Curling = 0,0824cm Top Tensile stress= 3,66 kg/cm2

140 cm x175 cm Thickness= 18 cm Delta T°=-14 C° Linear Curling = 0,049 cm Top Tensile stress= 1,83 kg/cm2

Table 6: Curling and stresses comparison between slabs of different lengths

Fig 7 .- Measured curling on an industrial floor. It shows that short slabs have

less curling than long slabs. (Holland 2002)

It has been seen that construction timing and curing are big contributors to curling of

concrete slabs, together with length (Table 6, Fig 7)

Sawing of joints has an important effect in curling and cracking. If sawing of the joints is

delayed curling is increased. It leaves the slabs long when the maximum differential

shrinkage is present. This can be an important cause of high upward curling, and

therefore longitudinal and transverse cracking. It is important to shorten the slabs as soon

as possible, before the surface force is produced. This can be accomplished utilizing

SoffCut or the thin wet saw (2 mm wide cut). Both can start sawing as soon as the

concrete slab can be walked over.

D) Effect of Load Transfer.-

When the slab length is reduced, bellow a given length, the stresses produced by traffic

loads change. For long slabs, load transfer helps in supporting the loading. For short

slabs, load transfer adds the loading of the adjacent slab and increases stresses. This is

shown in Fig. 10, where it can be seen that eliminating the load of the contiguous slab

reduces the stresses. This can also be seen in Fig 8, where the tie bars increase the

cantilever and the cracking of the slabs, by reducing the possibility for the slab to rock

and accommodate the loads in a less stressful position.

Fig 8.- Represents de effect of dowels and tie bars on the lengths on cantilevers

E) Stresses on Short Slabs.-

Normally, on 3.5 to 5 meters long slabs, the front and rear axels apply the load

simultaneously at both edges of the slab. This loading induces the traffic surface tensile

stresses to the top of the pavement when it is curled upwards, inducing top down cracks.

These tensile stresses at the top are due to the moment produced in the cantilever part of

the slab. This moment and stresses are smaller for shorter slabs, as the cantilever is

shorter. In this situation, it is very important the load transfer, which allows more than

one slab taking this loading.

If the slabs are short, of a length where the front and rear axels will never load the edges

simultaneously (Table 9), the configuration of the loading and the rocking of the slabs

change the stresses configuration within the slab. Only one set of wheels will move over

the slab and the slab will rock in a way that the load will always be touching the ground,

therefore well supported. In this case the slab will have no stresses produced by the

cantilever and the loading. In rocking, the slab will be lifted and the weight of the slab

will induce tensile stresses at the surface (Table 9). Now, the stresses are produced by the

slab’s own weight when it rocks. The main loading will depend on the geometry of the

slab and not on the traffic loading. If the slabs are curled upward and allowed to rock, the

stresses will be reduced, assuming the stiffness of the base is optimal.

Hiller and Springenschmid, 2004Hiller and Springenschmid, 2004

The following is a finite element model of the principal described above:

- Thickness: 20 cm of concrete

- Loads 1500 kg on each side

4,5m x 1m 2.25 m x 1 m

Maximun tensile Strength = 24.65 Kg/cm2 Maximun tensile Strength = 5.22 Kg/cm2

Principal stresses on the top of the slab, Red is tensile strength

Deformation of the slab

Table 9: Compararison between long and short slabs loaded by the same truck.

TCP Design Procedure:

A) Introduction

The concept behind the "TCP" design is that each slab of the pavement is loaded by only

one wheel or a set of wheels at the same time. This reduces significantly the top tensile

stresses of the slabs. With this configuration of loads versus the dimensions of the slab

the cantilever effect is reduced, so each slab supports the loads under the wheels,

supported on the ground.

B) Theory

To achieve this condition, it is necessary to dimension the slabs in such a way that given

a patron truck; each wheel, or a set of wheels, step a slab at the same time (Fig 10). As

different types of vehicles they exist, it is designed for the most harmful, unless the traffic

is known and it is designed for the vehicle type that will use the road

Fig 10:- Lateral view of the patron vehicle load

Reducing the tensions on the top of the slab allows a longer lifespan of the pavement, or

seen from another point of view, it allows a reduction of the thickness to achieve the

same tensions and lifespan obtained in the traditional design.

It is in the last case where the tensions generated by the loads are smaller, so the

pavements can be designed with thinner concrete slabs

The previous outline can be represented by the following graph (Table 11) of tensions in

the top of the slab, for different geometric configurations of the loads respect to their

dimensions. The model and use of this program are described later on in this work.

T

Thickness: 25 cm Concrete Slabs 4,5m x 3,6m

Thickness: 15 cm Concrete Slabs de 1,8m x 1,8m

Thickness: 13 cm Concrete Slabs 1,4m x 1,8m

Nota: Tension in red = traction; blue = Compression

Table 11:- Comparison between slab size optimized with the load geometry and the

thickness to achieve the same tensile stresses

Principal Stresses

23.50

22.14

20.10

18.06

16.02

13.98

11.95

9.91

7.87

5.83

3.79

1.75

-0.28

-2.32

-3.00

In the previous graphics is shown that the tensions are similar on the top of the pavement,

although the thickness diminishes considerably.

Graph 12: Tensions for different slab sizes.

Using ISLAB 2000, we can model different configurations of loads versus sizes and

geometries of slabs (Graph 12). It is important to highlight that smaller slabs don't always

have a better behavior.

C) Soil Support Model

Another important aspect in TCP design is the stiffness of the soil layers model. It has

always been considered that the concrete slab is sufficiently rigid to support the loads of

the vehicles. In the case of a thick slab (Over 14 cm) the base must should be soft to

support the slab with a bigger area(CBR 20%-50%). As the slab is supported in a bigger

area de cantilevers are smaller and the tensile stresses on top are reduced.

140x175 175x175 250x175 300x350 450x350

Tension Superior

12.99 18.64 15.78 27.6 29.82

0

5

10

15

20

25

30

35

Ten

sio

ns

g/cm

2

Principal Top Tensile Stresses

When using TCP design, the cantilevers are reduced, because the design optimizes the

size of the slab so this happens, the product is a thinner slabs in which the main tensile

stresses, are on the bottom of the slab just below the wheel, and a stiff base will reduce

them by collaborating with the slab to support the loads. It can be considered as semi-

rigid pavements (fig 13).

Fig 13: Comparison en the effect on bottom stresses on thin pavements

The previous graph shows thinner than 12 cm, this requires that the base has to be more

rigid to collaborates with the loads under the wheels that produce bottom tensile tensions.

It is for this reason that for these thicknesses the base should have an enough rigidity to

take these loads (CBR > 80%)

D) Design procedure

The way of designing this new concept in concrete pavements is being studied at Illinois

University, USA. For the time being, the design method being applied is the following:

– Design with AASHTO for 4.5 m long slabs

– Determine with finite element method (ISLAB 2000) the stresses in the slab, top

and bottom for these slabs designed with AASHTO.

– Calculate by iteration with ISLAB 2000 the thickness that gives the same stresses

on smaller slabs with the worst location of the wheels.

3 10 20 50 100

20 cm 15.1729 14.1132 13.5684 12.3986 11.3776

10 cm 48.349 41.9773 40.1801 37.0952 34.6339

0

10

20

30

40

50

60

Ten

sile

str

ess

es

bo

tto

m K

g/cm

2

CBR (%) Base

Soil Support Effect

20 cm

10 cm

E) New technology

The TCP design requires a great amount of saw cuts in the pavement, because of this, it is

necessary to use new technologies in its construction and maintenance. Within these

technologies are:

1. 2mm wide joints using thin saw blades to prevent spalling by uncompressibles in

the joint. These blades have been used in Chile from the year 2003.

2. Granular Base with less than 6% fine below sieve #200: Due to the quantity of

unsealed cuts, it is necessary to have a base whose structure is not affected by water.

When the granular base material has less than 6% fines, the particles of bigger size are in

touch with each other, so when the fines are washed out, volumetric change of the base

doesn't exist and since it is confined, there is no loss of support.

3. Impermeable or geotextile layer between Base and natural Soil: Natural soils

usually contain fines, because of this, it is fundamental to isolate the base from the

natural soil, to avoid transportation of fines to the base.

4. Lateral Bars for confinement: The TCP design considers, as part of the design, the

rocking of the slabs as the loads move over it. If the movement is restricted, the tensile

stresses are increased, reducing the life of the pavement. The amount of saw cuts made,

also make very difficult and expensive the use of dowel and tie bars. To solve this

problem, the pavement is confined on the exterior preventing the lateral displacement of

the slabs. This confinement consists on the placement of steel bars (16 mm) 50 cm long

buried vertically in both external sides of the pavement.

F) Calculations using ISLAB 2000

ISLAB2000 is a finite element program designed specifically to model concrete

pavements.

The parameters of used in the model are:

1. Properties of the concrete:

1E = 290.000 2cm

Kg

= 0,25

= 0,0025 3cm

Kg

= 1 x 510 C

1

2. Soil layers model

The soil layers are modeled as a spring with a k value. To determine this value the

procedure used is described in: "Pavement Design according to the AASHTO method"

1993 p.82" to correlate k values with CBR.

3. Position of the loads:

To carry out the design, the critical position of the loads is considered. In the longitudinal

way the worst condition is on the edge of the slab.

In the traverse direction, it was evaluated in three positions: wheel path, Border and 15

cm from the border, in the case of designing the pavement with widened outer lane.

4. Vehicle considered in the analysis:

The selected truck to do the modeling is a Mercedes Benz truck model LK2638 6x4

whose specifications were obtained from the representative in Chile, and are:

Distance between wheels, same

longitudinal axle (cm)

Distance between

wheels, same transverse axle

(cm)

Distance between front and rear axle (cm)

Type of Axle

D1

D2

D3

ESRS

-

-

468

EDRD

145

6

Table 14 –Mercedes Benz Model LK2638 6x4

The loads of the axes are considered as the standard axel in Chile (7 tons, 11 tons and 18

tons; ESRS, ESRD, EDRD respectively) plus 20 overload%. The tire `pressure used was

8 kg/cm2 (120 psi)

5. Thermal Gradient:

To evaluate the different conditions, the model considers equivalent warping and curling

to thermal gradients between +5 °C and -21 °C.

The construction curling is equivalent to a thermal gradient of -11°C, which should be

added to the normal thermal gradients.

To carry out the comparisons in this work, it was considered, as the worst case, an

equivalent curling to a thermal gradient of -17 °C.

Summary and Conclusions

With short slabs and half a lane wide, the design concepts change. With this geometry the

stresses are mainly due to the own weight of the slab and the stresses under the wheel. .

The short slabs curl much less than longer slabs. Allowing the rocking of the slabs should

reduce stresses in the pavement. Therefore, tie bars and load transfer should not exist.

The design does not include steel bars within the slabs. Lateral confinement, to eliminate

a possible drift and separation of the lanes, can be achieved with curbs or by vertical steel

pins on the outer edges of the pavement slabs as curling is low, faulting due to traffic

should be reduced.

These shorter slabs require more joints and saw cuts. To reduce the cost of this type of

construction, the sawing should be made with a thin saw leaving a cut 2 mm wide, where

incompressible particles don’t get in. No sealing is needed, and to prevent damage from

the water penetration, a cement or asphalt treated base or a granular base with less than

6% fines. This last material, more than being drainable, doesn’t de-compact with

pumping, as the fines are only filling the voids of the bigger aggregate. If the fines are

taken away, there would be no volume change in the base.

An important point is that the sawing of the joints should be done before the construction

curling occurs. This is as soon as the slab can be walked over to start the sawing

operation.

To check the behavior of the TCP design, more than 100.000 runs with ISLAB 2000 have

been executed, obtaining solutions optimized for most of the combinations of the design

variables. Due to the quantity of information obtained, design software is being

developed to facilitate the use of this technology. Parallel to this, a study is being carried

out at the University of Illinois, USA.

The TCP design allows a reduction in thickness between 6 and 10 cm compared to

concrete pavements designed by the AASHTO 98 method, reducing the construction cost

in about US $ 25.000 per kilometer-lane, corresponding to about 20% of the initial cost,

with a similar life. When comparing with asphalt solutions, the savings are similar.

The next graph shows an analysis of initial cost, for equivalent solutions, in asphalt and

in concrete, with values used by the Ministry of Housing and Urban Development of

Chile.

Fig 15 : Economical comparison between asphalt and concrete solution in initial cost

Because they don't have joint seals, the maintenance cost is lower

Experience:

In Chile test tracts have been constructed and some urbanizations contracts have

been executed with very good results. Their behavior is considered superior to the

projected one.

In Guatemala, from 2005 up to now, there are 100 km of 4 lanes highways

designed and constructed with this system. The design is for a high traffic (50.000.000

EE), with thickness of 17 cm of traditional concrete.

The behaviors have been considered very good by the Ministry of Communications of

this country.

If you want to know more about this technology you can visit:

www.tcpavements.com

REFERENCES

-Yoder E.J. and Witczak M.W., Principles of Pavement Design, Second Edition, John Wiley & Sons, Inc., 1975.

-AASHTO Guide for Design of Pavement Structures, American

Association of State Highway and Transportation Officials, USA, 1993. -Larrain, C. Análisis Teórico-Experimental del Comportamiento de

Losas de Hormigón de Pavimentos, MSc Thesis, School of Engineering, Catholic University of Chile, 1985, 280 pp.

- Análisis del Comportamiento de Pavimentos Delgados en Condiciones de Carga Pesada. Modelación con Software de Elementos

Finitos ISLAB 2000, Enero, 2006. Juan Pablo Covarrubias Torres, Daniel Andahur.

-Eres Consultants, ISLAB 2000. Finite Element Program for the Analysis of Rigid Pavements, Version 1.1, USA, 1999.

- HDM 4 Series, International Study of Highway Development and

Management Systems, 1997. - Hiller and Springenschmid, “The Influence of the Curing Method on

Early Cracking Risk During Hot Weather Paving” 9th Internacional

Symposium on Concrete Roads, 04-07 April 2004, Istanbul, Turkey.

TCP technology (Thin Concrete Pavements), the methodology for the design and construction of improved concrete pavement slabs and other rights related to this technology (software, know-how, industrial secrets, trademarks, manuals, instructions, etc.), are exclusively owned by Comercial TCPavements Ltda. and subject to legal protection as recognized in local regulations and International Intellectual and Industrial Property Treaties, particularly by patent applications Nos. 2684-05 in Chile, international application PCT/EP2006/064732, application 20070094990 in the United States. ©TCPavements 2005-2007, all rights reserved.