STATISTICAL MODEL FOR EFFECT OF

POLYMER MODIFIED ASPHALT ON RUTTING

POTENTIAL OF ASPHALT CONCRETE ROAD

DEPARTMENT OF CIVIL ENGINEERING

NED UNIVERSITY OF ENGINEERING AND TECHNOLOGY

PAKISTAN
 STATISTICAL MODEL FOR EFFECT OF
POLYMER MODIFIED ASPHALT ON RUTTING
POTENTIAL OF ASPHALT CONCRETE ROAD

By

Sajjad Ali
Seat No. CE -053
Batch 2011-2012
In partial fulfillment of requirements for Master’s Degree, in Civil Engineering

DEPARTMENT OF CIVIL ENGINEERING

NED UNIVERSITY OF ENGINEERING AND TECHNOLOGY

ii
 CERTIFICATE

This is to certify that Mr. Sajjad Ali, student of BATCH 2011-2012, Bearing Roll No.
CE–053 has successfully completed the Independent Study Project (ISP), equivalent to
six credit hours, in partial fulfillment of the requirements for a Master’s degree in Civil
Engineering, NED University of Engineering and Technology.

PROJECT SUPERVISOR CHAIRMAN

__________________________ _________________________
Prof. Dr. Adnan Qadir Prof. Dr. Asad ur Rehman Khan
Professor Professor\Chairman
Department of Urban &Infrastructure Department of Civil Engineering,
Engineering NED University of Engineering &
NED University of Engineering & Technology.
Technology.

iii
 TABLE OF CONTENT

LIST OF TABLES vi

LIST OF FIGURES vii

NOTATIONS AND ABBREVIATIONS viii

ACKNOWLEDGEMENT ix

ABSTRACT x

Chapter 1 1

Introduction 1

1.1. General 1

1.2. Objective 4

1.3. Scope and Limitations 5

1.4. Independent Study Details 5

Chapter 2 6

Litreture Review 6

Chapter 3 9

Research Methodlogy 9

3.1. Materials Used in the Study 9

3.2. Aggregate 10

3.3. Polypropylene Fiber 12

3.4. Properties of Base Asphalt and Modified Asphalts 12

3.5. Rutting Sample Mould and Size 14

3.6. OAC of the Mixture 15

3.7. Rutting Susceptibility Test 15

Chapter 4 17

RESULTS AND DISCUSSIONS 17

4.1. Introduction 17

4.2. Rut Depth (Constant Temperature & Variable Polymertype) 17

4.3. Rut Depth (Same Binder Type & Variable Temperatures) 19

4.4. Summary of Graphs 21

iv
Chapter 5 23

Model preparation and validation 23

5.1. Model Preparation 23

5.2. Coding for Empirical Modeling 23

5.3. Design of Experiment Matrix 24

5.4. Regression Equation Selection And Validation 25

5.5. Analysis of Variance 25

Chapter 6 27

Conclusions & recommendations 27

6.1. Conclusions 27

6.2. Recommendations 27

References 28

Annexure 30

v
 LIST OF TABLES

Table 3.1 Test Methods& Results of Aggregates 10

Table 3.2 Particle Size Distribution 10

Table 3.3 Physical Properties of Polypropylene Fiber 12

Table 3.4 Physical Properties of Base Asphalt 13

Table 3.5 Physical Properties of Modified Asphalt 13

Table 3.6 Optimum Asphalt Content 15

Table 4.1 Rutting Results 22

Table 5.1 Notation for the Regression Equation 23

Table 5.2 Design of Experiment 24

vi
 LIST OF FIGURES

Figure 3.1 Gradation Curve of Aggregates Used in this Study 11

Figure 3.2 Mixing of PP Fiber in Asphlat 14

Figure 3.3 Sample Prepared for Rutting Test 15

Figure 3.4 Roller Compacter for Compacting Specimens 16

Figure 3.5 Wheel Tracking Device for Rutting Susceptibility Test 16

Figure 4.1 Rut Depths at 50°C 17

Figure 4.2 Rut Depths at 55°C 18

Figure 4.3 Rut Depths at 60°C 18

Figure 4.4 Rut depths of Neat Asphalt at variable temperatures 19

Figure 4.5 Rut depths of Polypropylene fiber at variable temperatures 20

Figure 4.6 Rut depths of polymer Modified Asphalt at variable temperatures 20

Figure 4.7 Summary Graph 21

Figure 5.1 Model Validation Curve 26

vii
 NOTATIONS AND ABBREVIATIONS

AC Asphalt Concrete

OAC Optimum Asphalt Content

ASTM American Society for Testing and Materials

EN European Standard

RCA Recycled Concrete Aggregates

RAP Recycled Asphalt Pavement

HMA Hot Mix Asphalt

PPF Polypropylene Fiber

PMA Polymer Modified Asphalt

NHA National Highway Authority

WTD Wheel Tracking Device

PT Polymer Type

RC Roller Compactor

viii
 ACKNOWLEDGEMENT

In the name of Allah the most beneficial, the most compassionate. By the grace of Him, I
have successfully completed this Independent Study Project (ISP). The Project bearing
the title “STATISTICAL MODEL FOR EFFECT OF POLYMER MODIFIED
ASPHALT ON RUTTING POTENTIAL OF ASPHALT CONCRETE ROAD” has been
supervised by our respected and hardworking Prof. Dr. Adnan Qadir. This ISP led me a
chance to prove myself up to the expectation of the Professor and the requirement of the
Department. During the tiring work of the experiments for the ISP opens new door of
patience along with the enhancement of faith on me for the completion of the work. The
literature review reveals me to the new horizon of the science more specifically the
flexible pavement.

I would like to thank Prof. Dr. Adnan Qadir for their support and guidance throughout the
study, his guidance directs me to complete the study. I would also like to thank Prof. Dr.
Mir Shabbar Ali, Chairman Department of Urban and Infrastructure Engineering, Prof.
Dr. Masood Rafi, Chairman Department of Earthquake Engineering and my family
member for their care and support. I am pleased to my colleagues and friends as well.

ix
 ABSTRACT

This study investigates the use of neat asphalt, polymer modified asphalt and poly
propylene fiber in asphalt concrete for rutting susceptibility. Rutting is one of the
common distress measures for all type of asphalt concrete mixes. Test are conducted at
different temperature levels of 50°C , 55°C and 60°C.The sample is prepared on optimum
asphalt content (OAC). Optimum asphalt content is obtained by Marshall Mix design
method. Polypropylene fibers are mixed in Hot Mix Asphalt using the “wet method” i.e.
the fibers are mixed and heated with asphalt binder before being added to aggregates.
Polypropylene fibers are added 0.5% by weight of aggregates. In polymer modified
asphalt (PMA), SBS is mixed before testing. Rutting susceptibility tests are carried out on
above mentioned three temperature levels. For each temperature levels and asphalt mix
five sample are tested. A total of forty five samples are tested .Increase in Rut depths are
observed as temperature increase. Reduction in rut depth was observed for the modified
samples. PMA has greatest Rut resistance among all while neat asphalt mix has the
lowest rut resistance.

Sample is prepared for rutting in Roller Compactor device. The sample is then cured for a
day or minimum overnight. Rutting test is performed on Wheel Tracking Device.
Temperatures for tests are 50°C, 55°C and 60°C as stated before. Number of passes kept
constant at ten thousands passes (10,000 passes & 5000 load cycle). Design of experiment
is then prepared for statistical Model; Minitab software is used for the statistical analysis.
The regression equation is then produced. Equation contains Rut Depth (R.D) as
dependent variable having parameters of temperature (T) and Modifier type (P).

x
 DEDICATION

This research is dedicated to my parents especially my mother, Her endless blessing and
mercy made me able to think and perform positive, along with it is dedicated my family
members. At last it is also dedicated to my teacher and specially Prof. Dr. Adnan Qadir.

xi
 Chapter 1

INTRODUCTION

1.1. GENERAL

Roads in Pakistan are in poor condition. The distresses like rutting and fatigue cracking
have degraded the condition of the available road network making them unsafe and
dangerous for the road users. Since road transportation complements the other modes of
transportation, the poor condition of roads in Pakistan is affecting the country’s
economy. Currently Pakistan has almost 260,000 km of road network which is widely
used and carries 91% passenger traffic and almost 96% of the freight traffic[1].Hence it
is vital to keep the roads in good condition in order to facilitate smooth and safe traffic
flow over the road facilities. Different reasons can be cited for the deteriorating
condition of roads including violation of maximum permissible load carrying rules for
transport vehicles and shortage of funds for maintenance activities. Unfortunately, the
gap between the funds required and available for maintenance and rehabilitation
activities has never been bridged since 1999-2000 [2].In this scenario it is crucial to
come up with a pavement design that is durable and more resistant to distresses. Such
pavements will have greater service life than the conventional pavements and will,
therefore, require lesser maintenance. One method of making durable pavements is by
modification in pavement materials. In this study the addition of polymers and
elastomers are investigated at different temperature levels. Conclusively, it can be said
that this study presents an empirical model for modification in the pavement materials
at different elevated temperatures level. The results are based on the laboratory
evaluation of material behavior on the basis of modification in the pavements materials
to reduce rutting.

The distresses in the pavements are varied and complex to understand. The primary
reason making the distresses difficult to understand is the bitumen, the binder in the
wearing course, has highly unpredictable behavior. It behaves differently at different
temperatures and under different loading conditions. The deterioration of the pavement
restricts it from achieving its full service life. Rutting is one of the most common
distresses encountered by the pavement. It results from repetitive passes of high axle
load vehicles. Rutting is permanent deformation of the wearing course which appears
along the length of road in the wheel path. The ruts become evident after rainfall when
filled with water. Rutting is dangerous as it causes hydroplaning making the vehicle
1
lose its ability to brake and control. Also the ruts in the road steers the vehicle towards
the rut path. Hence rutting is not only a nuisance but also dangerous for the vehicles.

Different solutions have been proposed and used for control of rutting in flexible
pavements. One such solution is the use of polymer fibers which is widely employed all
around the world. Polypropylene is very effectively used for controlling rutting.
However study needs to be carried out to investigate the effectiveness of polypropylene
fibers in our local conditions using the local aggregates. Locally tested pavement design
method incorporating the use of polypropylene (PP) fibers would take care of the
distresses in the pavements in the local conditions. The study would comprise of
laboratory testing of polypropylene by comparing with the conventional Hot Mix
Asphalt (HMA) samples. This would not only ensure in an asphalt pavement that is
more resistant to rutting than the conventional pavements but former would be more
durable and long-lasting. On the other hand the SBS modified asphalt is also one such
technique to reduce rutting .SBS elastomer is also incorporated to grabs the said
problem. SBS modification is common all around the globe. The study aims to test it for
the local conditions and made it convenient to use in the local environment. The study
will compare it with HMA and made an argument for it. So, at a glance the study aims
to the empirical relation of rut depth, temperature variation and addition of elastomer
and polymers in HMA. It is then compare with the neat asphalt mix.

Asphalt is quite an unpredictable material in the asphalt concrete pavements. It behaves

differently under different loading conditions and at different temperatures. Asphalt is
viscoelastic material that sometimes behaves as viscoelastic-plastic material at different
temperatures. Therefore sound knowledge of asphalt behavior is necessary in order to
find the solution to the various distresses in the flexible pavements.

Rutting is one of the most common distresses. It is permanent deformation which

appears as longitudinal cracks occurring in pavement along the wheel path. It is the
severe distress which usually generates due to repeated loading and effect serviceability
of pavement. Shearing and pavement uplift occurs along the side of rut.

Rutting can be further classified into mix rutting and subgrade rutting. In mix rutting the
subgrade doesn’t rut and the permanent deformation takes place in the pavement layers.
While wheel path depressions are formed in subgrade and pavement settles into the ruts
causing surface depressions in the case of subgrade rutting.

2
The general cause of rutting is due to the displacement or lateral movement of the
material particles in any pavement layer directly under the load or in the wheel path.
Specific reasons may include:

i) Insufficient compaction of HMA layers or presence of excessive air

voids during construction. If it is not compacted enough initially, HMA
pavement may continue to consolidate under traffic loads.
ii) Subgrade rutting due to lack of proper strength or inadequate pavement
structure.
iii) Improper mix design or manufacture (e.g., high asphalt content,
excessive mineral filler, insufficient amount of angular aggregate particles).
iv) Hot climate may cause the softening of asphaltic binder which will result
eventually in rutting.
v) Asphalt binder properties influence the permanent deformation
resistance properties of asphalt mixtures. The stiffer the binder the greater will
be the resistance to deform.

In Pakistan, the roadways are usually made as flexible due to low cost as compared to
rigid pavements. Flexible pavement experiences traffic loading and environmental
factors (rain fall and temperature) that affect the performance of pavement and creates
various distresses [3].According to Pakistan Bureau of Statistics report, 2011 the total
road network of Pakistan is 259,643 km, most of them in depleted conditions.
According to National Highway Authority (NHA) 67% of their road network is in poor
condition with rutting being the most common form of roads distresses.

Asphalt modification by addition of different materials in the recent times is a very

successful and effective technique for improving its rheological properties.
Modification of asphalt by the addition of polymers and elastomers are solution to
various distresses of the pavement. It is considered as one of the solution to improve
fatigue life, reduce rutting and thermal cracking in the pavement. [4]Incorporating
materials like fibers and SBS in the pavements is innovative way of increasing its
design service life. The increased service life of pavements provides durable roads
which require lesser maintenance than the conventional asphalt concrete pavements.
Use of Polypropylene fibers in asphalt concrete has been widely adopted all around the
world in this regards. However, no detailed study has yet been made investigating the
use of these fibers in pavements in Pakistan under the local traffic loading and
3
environmental conditions. This research study can be regarded as the stepping stone for
the future studies that could be carried out for investigating the numerous parameters
affected by inclusion of Polypropylene fibers in pavement. The study for the addition of
SBS are been carried out by many researcher but yet not able to develop a through
guidelines for using it in commercial environment in developing countries like Pakistan
and many others. The study aims to have step towards the effectiveness of modification
in asphalt for the betterment of the country in nut shell and for the society at large.

This research has aimed to investigate the use of Polypropylene fibers and elastomer in
asphalt concrete of pavement. Rutting in the asphalt course being one of the most
common distresses, responsible for reducing the design service life of pavement, has
been investigated by inclusion of fibers in asphalt concrete. Polypropylene fibers were
mixed in Hot Mix Asphalt using the “wet method” i.e. the fibers were mixed and heated
with asphalt binder before being added to aggregates. Polypropylene fibers were added
0.5% by weight of aggregates. Marshall Mix Design was used to determine the
Optimum Asphalt Content (OAC) of HMA incorporating fibers. The OAC was
determined for both control (samples without fiber) and modified (samples containing
fibers). Along with it the polymer modified asphalt has also check for OAC .by getting
the OAC of the thrice i.e. neat asphalt, polypropylene and PMA.it is then use to prepare
the rutting samples. The rutting samples are then tested at 50°C, 55ºC and 60ºC... At
low temperatures i.e. around 40ºC, the inclusion of Polypropylene fibers increased the
air voids content leading to a greater rut depth than the control samples. The fibrous
pavements are performing well at elevated temperatures as compare to the low
temperature, while PMA has better result than Polypropylene in both temperatures.

1.2. OBJECTIVE

Aim of the study is to obtain the following objectives;

 To evaluate the rutting potential of asphalt concrete mix with the addition of
polymers and elastomers.
 To know how these polymerization help helps in sustaining rutting at elevated
temperature.
 To come up with empirical model of polymerization of asphalt concrete.

4
1.3. SCOPE AND LIMITATIONS

 The rutting test will be limited only to the investigation of mix rutting.
 The axle load for the rutting test which will be limited to 13 kips
 The proposed mix design is expected, but not guaranteed, to perform
satisfactorily under axle load greater than 13kips.
 Also the testing temperature for the rutting test is restricted to 60°C. The rutting
tests cannot be conducted at temperatures greater than 60°C since this is the
maximum temperature achievable in the Wheel Tracking Device.
 Hence, the proposed mix design proposed may not be applicable for field
conditions with temperatures over 60°C.

1.4. INDEPENDENT STUDY DETAILS

The Independent study project is designed in the following sequence;

Chapter 1 discusses the introduction of the study its importance and need to the
pavement industry along with the daily human life. Chapter 2 includes the literature
review regarding the use of recycled aggregates and Polymer Modified Asphalt (PMA)
at different temperatures and loading conditions. Chapter 3 comprises of testing
methodology of the study. It covers the brief description of the procedure adopted for
this study. Chapter 4 discusses the results obtained from this study. Chapter comprises
of empirical modelling technique and development of model. It also includes the model
validation of the study. Chapter 6 provides the conclusions and recommendation based
on the results.

5
 Chapter 2

LITERATURE REVIEW

The effect of Polypropylene in Hot Mix Asphalt has been investigated throughout the
world in detail. This section provides brief reviews of researches undertaken all around
the world in this regards. Based on these researches, the methodology for this research
project was formulated.

For designing a rut-resistant pavement, the mechanism of rutting should be understood

completely. Two different mechanisms leading to rut development has been discussed
by [5]. The first one is associated with well-designed structures and stable materials.
The permanent deformation is due to the effect of repeated load on pavement materials
that are not perfectly elastic. The amount of rutting depends on the stiffness, the
permanent deformation potential of the different materials and the load induced stresses.
In the second mechanism, the rut is formed due to the failure of the Asphalt Concrete
(AC) layer. This is a localized process induced by the large shear stress near the surface
in an unstable material. This problem needs to be addressed by testing and construction
specifications. Note that in well-designed structures with a stable AC material, a shear
stress of similar magnitude develops near the pavement surface; however, it does not
cause the failure of the materials.

Improvement of rheological properties of asphalt binder can be achieved through

different methods. Bitumen modification is a prevalent and effective means of
achieving this goal. Commonly polymer fibers have been used for modification of the
binder. Work on polymer modification for the reinforcement of asphalt concrete began
in the early 1990s. Brown et al [6] have worked greatly for the development of polymer
fiber reinforcement and composite materials. The research has identified the potential of
specific fibers for improving the tensile and cohesive strength of asphaltic concrete due
to developing greater tensile strength when compared to bitumen.

Jew and Wood hams [7] found Polymer-modified binders (PMBs) to improve several
properties of paving mixtures, such as temperature susceptibility, fatigue life, and
resistance to permanent deformation. The polymers generally used for the modification
of asphalt cement for paving purposes are styrene-butadiene-styrene copolymer,
styrene-butadiene-rubber, latex, ethylene vinyl acetate copolymer, low-density
polyethylene (LDPE), or linear low density polyethylene and polypropylene.

6
Polypropylene fibers have been in similar use for quite some time now. These fibers
have been particularly used for the purpose of controlling the permanent deformation,
rutting [8] also the pavements have been found to be more durable and have greater
service-life after polymer modification because the polymer addition was found to
improve the ductility of the binder [9].

Marshall testing has shown increased stability and flow values by addition of
polypropylene fibers. Ebrahimi [10] prepared asphalt specimens by Superpave Gyratory
Compactor (SGC). These samples were analyzed by both Marshall Analysis and
Superpave Analysis and finally tested by Marshall Stability. It was observed that adding
PP showed increasing in Marshall Stability (26.3%), percent of air void (67.5%), and
also decreasing Flow (38%). These results show increase pavement life service, also
increasing % air void is useful for hot regions which bleeding and flushing are
important distresses.

Tapkin et al [11] performed repeated creep tests and concluded the addition of the
polypropylene fibers into the asphalt mixture increased the Marshall Stability value by
20%. It can be concluded that the lives of the fiber modified asphalt specimens under
repeated creep loading at different loading patterns increased by 5–12 times versus
control specimens. This is a significant improvement. The results from the analysis of
the tested specimens show that the addition of polypropylene fibers improves the
behavior of the specimens by increasing the life of samples under repeated creep
testing.

Kamil et al [12] used a mixture of polypropylene and aramid fibers in a field and
laboratory study to evaluate the performance characteristics of the modified asphalt
mixture. The laboratory experimental program on the field mixes included: tri-axial
shear strength, dynamic (complex) modulus, repeated load permanent deformation,
beam fatigue, crack propagation, and indirect diametral tensile tests. The results showed
that the fibers improved the mixture’s performance in several unique ways as
summarized below:

• The fiber-reinforced asphalt mixture showed better resistance to shear

deformation.

• The measured Dynamic Modulus E* values were higher for the fiber-reinforced
mix.

7
 • The tensile strength and fracture energy measured from the indirect diametral
tensile tests showed that at all test temperatures, the fiber-reinforced mix exhibited
the highest values.

• Fiber-reinforced mix had about 40 times higher resistance to crack propagation

than the control mix.

• A field condition survey after approximately one year (with two summer periods
included) revealed that there are a couple of low severity cracks, 1 to 2 feet long,
in the control section. No cracks were observed in the fiber-reinforced pavement
sections.

Simpson and Kamyar[13] conducted another study in which polypropylene, polyester

fibers, and some other polymers were used to modify the bituminous binder. The testing
procedures included Marshall Stability, IDT, moisture damage susceptibility,
freeze/thaw susceptibility, resilient modulus, and repeated load deformation. Mixtures
containing polypropylene fibers were found to have higher tensile strengths and
resistance to cracking. Rutting potential as measured by repeated load deformation
testing was found to decrease only in polypropylene modified samples. Hence the
superiority of using polypropylene over the conventional methods was established in
this research.

From the literature it can be concluded that the propylene mixed asphalt concrete is
behaving well under different testing conditions. A number of researchers found the
propylene modified asphalt concrete rut resistant at various loadings. However the
rutting behavior of such concrete at different temperature is yet to be explored.
Moreover the authors of this report failed to get any literature evident of such study ever
taken place in Pakistan. Therefore based on the studied literature, the following
objectives for this study were derived.

8
 Chapter 3

RESEARCH METHODLOGY

The research methodology of this study included rutting susceptibility of neat asphalt
sample, the sample made using PMA and polypropylene samples. Each type of sample
made on OAC determined by Marshall Mix Design Method. Each sample has it
different OAC. The samples were then tested at different temperatures of 50°C, 55°C
and 60°C. For each set up of experiments five sample were tested. The finally an
empirical relation were determined to have a generalize equation for the both modifier
and neat asphalt against the rut depth.

The Marshall samples were prepared using five (5) levels of addition of asphalt content
i.e. 3.5%, 4.0%, 4.5%, 5.0% and 5.5%. A total of fifty (75) Marshall Samples were
prepared in which twenty five (25) samples were made for each category i.e. neat
asphalt mix, PMA mix and Polypropylene mix respectively. Polypropylene was added
as 0.5% by weight of aggregates [14]. After determining the OAC, samples for rutting
were prepared. A total of forty five (45) samples were prepared for rutting tests. A total
of fifteen (15) samples were made with neat Asphalt and remaining thirty (30) was
distributed equally between PMA mix and polypropylene mix. The distribution of
fifteen rut samples of each category is bases on temperature. At temperature of 50°C
five (05) samples were prepared and the rest of ten (10) were equally distributed on
temperature of 55°C and 60°C respectively. These samples were then tested in Wheel
Tracking Device. The rut depths obtained from rutting tests of samples made with
polypropylene and PMA were later compared with the rut depths of control samples to
determine the effectiveness of modification in increasing the rut resistance. Empirical
models that show the relationship between rut depths, modification of materials and
temperature was also been develop.

3.1. MATERIALS USED IN THE STUDY

This section gives the detail of all the materials used in this research. There were three
different types of materials namely crushed aggregates, Polypropylene fiber and
Polymer modified Asphalt (PMA). Basic properties of all these material described
below. The Rut depth will not only provides the empirical analysis on the basis of
temperature and modification but the study can also fine the effect of modification
determines the effectiveness of modification in in mix design. Rutting susceptibility test

9
of asphalt paving mixture is done using Wheel Tracking Test Device were the main
objective with different temperature level as discussed earlier.

3.2. AGGREGATE

Aggregate is a collective term for the mineral materials such as sand, gravel and
crushed stone that are used with a binding medium (such as water, Asphalt, Portland
cement, lime, etc.) to form compound materials (such as Asphalt concrete and Portland
cement concrete). By volume, aggregate generally accounts for 92 to 96 percent of
asphalt concrete and about 70 to 80 percent of Portland cement concrete. The source of
crushed aggregates used in this study was brought from the quarry located at Hub
Baluchistan (latitude 25010210 and longitude 6701013) and have got the following
properties.

Table 3.1 Test Methods& Results of Aggregates

Method Title TEST METHOD Values
Specific gravity of aggregates – Coarse ASTM C127 2.681
Specific gravity of aggregates- Fine ASTM C128 2.678
Specific gravity of aggregates- Filler ASTM C 188 2.493
LA Abrasion, % ASTM C131 23
Impact value, % BS 812-112 18
Crushing Value,% BS 812-114 21

The aggregate gradation employed in this research is NHA Type A (Minimum). The
NHA’s gradation is shown in Table below
Table 3.2 Particle Size Distribution
Sieve Designation Percent Passing by Weight
mm Inch / No. A B
25 1” 100 -
19 ¾ 100
12.5 ½ - 75-90
9.5 3/8 56-70 60-80
4.75 No. 4 35-50 40-60
3.38 No. 8 23-35 20-40
1.18 No. 16 5-12 5-15
0.075 No. 200 2-8 3-8

Aggregate gradation curve is shown in Figure above. The particle size distribution, or
gradation, of an aggregate is one of the most influential aggregate characteristics in

10
determining how it will perform as a pavement material. In HMA, gradation helps
determining almost every important property including stiffness, stability, durability,
permeability, workability, fatigue resistance, frictional resistance and resistance to
moisture damage .Because of this, gradation is a primary concern in asphalt concrete
and thus most agencies specify allowable aggregate gradations. The aggregate gradation
employed in this research was NHA Type B (Central limit). The NHA’s gradation was
shown in Table 2 while the aggregate gradation curve is illustrated in Figure 2.

Sieve Sizes
(mm)
25.419 9.51 4.76 2.38 1.19 0.074
100

90

80

70 NHA Fine
Percentage Passing (%)

Gradation
60

50 NHA
Selected
gradation for
40
this study

30

20

10

0
1"3/4" 3/8" No.4 No. 8No. 16 No. 200
Sieve Sizes
(inch/No.)

Figure 3.1 Gradation Curve of Aggregates Used in this Study

11
3.3. POLYPROPYLENE FIBER

The type of Polypropylene fiber used in this research has been Monofilament
Polypropylene fibers. The different physical properties of this fiber are given in Table
3.2.

Table 3.3 Physical Properties of Polypropylene Fiber

Characteristic Value Standard
Color Transparent
Length (mm) 13
Compressive strength (psi) 5,500-8,000 ASTM D695
Flexural strength (psi) 6,000-8,000 ASTM D790
Tensile strength at break (psi) 4,500-6,000 ASTM D638
Elongation at break (%) 100-600 ASTM D638
Water Absorption (%) Negligible (0.01-0.03) ASTM D-570
Specific Gravity 0.9-0.91 ASTM D-792
Ignition Point 593°C
Melting Point 160 – 170°C
Heat and UV Stabilization Long term
Thermal conductivity 2.810-4cal cm/sec cm
Tensile modulus (ksi) 165-225 ASTM D-638
Compressive modulus (ksi) 150-300 ASTM D-695
Flexural Modulus (ksi @ 25°C) 170-250 ASTM D790
Rockwell Hardness R80-R102 ASTM D785
Electrical Conductivity Low
Salt resistance High
Acid Resistance High
Alkali resistance 100% (alkali proof)

3.4. PROPERTIES OF BASE ASPHALT AND MODIFIED

ASPHALTS
Asphalt (according to ASTM) is a black to dark brown solid or semi-solid cementations
material which gradually liquefies when heated. The predominating constituents are
bitumen, all of which occur in the solid or semi-solid form in nature or are obtained by
refining petroleum or which are combinations of the bitumen mentioned or with

12
petroleum or derivatives thereof. The properties of neat asphalt used in the study have
the following properties.

Table 3.4 Physical Properties of Base Asphalt

TEST METHOD METHOD TITLE Base Asphalt
BS 4692 /ASTM D-36 Softening Point Ring & Ball apparatus 46,47
Average Penetration at (25 ºC, 100 gm., 5
BS 4691/ASTM D-5 sec.) 98
ASTM D-113 Ductility at (25oC, 5 cm/min) 100
ASTM D-92 Flash point °C 344
ASTM D-445 Kinematic Viscosity @80°C, cst 13290
ASTM D-445 Kinematic Viscosity @100°C,cst 2388
ASTM D-445 Kinematic Viscosity @120°C,cst 819

The polymer additives do not chemically combine or change the chemical nature of the
bitumen being modified, apart from being present in and throughout the bitumen. What
polymers will do is change the physical nature of bitumen, and they are able to modify
such physical properties as the softening point and the brittleness of the bitumen. Elastic
recovery/ductility can also be improved. The base asphalt and PMA used in this study
have the following properties.

Table 3.5 Physical Properties of Modified Asphalt

TEST METHOD METHOD TITLE Modified Asphalt
BS 4692 /ASTM D-
Softening Point Ring & Ball apparatus 56,57
36
Average Penetration at (25 ºC, 100 gm., 5
BS 4691/ASTM D-5 64
sec.)
ASTM D-113 Ductility at (25oC, 5 cm/min) 100
ASTM D-92 Flash point °C 300
ASTM D-445 Kinematic Viscosity @80°C, cst 72483
ASTM D-445 Kinematic Viscosity @100°C,cst 6050
ASTM D-445 Kinematic Viscosity @120°C,cst 2039

On the other hand in order to mix PP fiber is a unique work to perform. A mix
technique called “Wet Method” [15] has been adopted in this research. The required

13
quantity of fiber (0.5% by weight of mix) was first taken in the pan and then heated
asphalt was added and mixed thoroughly until the mix acquires uniformity.

Figure 3.2 Mixing of PP Fiber in Asphlat

Once the asphalt propylene mixture was obtained then it was added with aggregates and
the mix was thoroughly mixed until each aggregate particle was adequately coated with
mixture. The prepared mix was further kept in oven at 120°C for four (4) hours for
aging of the mix. This is done to simulate short term aging normally associated in
preparations and laying of HMA in the field. The samples were later compacted in a
cylindrical mold of 4 inch diameter, by application of 75 hammer blows on each side.

3.5. RUTTING SAMPLE MOULD AND SIZE

Mold of rutting sample is 300 x 300 mm having depths of 25mm, 50mm, 75 mm, and
100mm. the mold selected for the study is 300 x 300 x 25 (all units are in mm). It is a
hollow square mold. The mold selected for the study is 300 x 300 x 25 (all units are in
mm). It is a hollow square mold in which you place asphalt concrete mixture. The mold
is then place in a roller compactor. After compaction and curing it is then ready for
rutting test.

14
 Figure 3.3 Sample Prepared for Rutting Test
3.6. OAC OF THE MIXTURE

The OAC of control as well as modified samples has been determined using the
Marshall Mix Design Method. After determining the OAC samples for rutting tests
were prepared. Density at OAC from Marshall Tests was used to calculate the mass of
aggregates required for rutting tests for both control and modified samples.

Table 3.6 Optimum Asphalt Content

Asphalt Mix Type OAC (%)
Neat Asphalt 4.8
PMA 6.5
PP Fiber 5.5

3.7. RUTTING SUSCEPTIBILITY TEST

The mixing of sample for rutting tests were done identically in the same manner as it
was for Marshal Test; however the samples were compacted in Roller Compactor
shown in Figure 3.4. Fifteen samples each of control; PMA and polypropylene were
thus prepared.

15
 Figure 3.4 Roller Compacter for Compacting Specimens
After compaction, the samples were tested in Wheel Tracking Device shown at different
temperatures level. Fifteen of each sample category and five at each temperature levels.
Average rut depth values were calculated for the three type of sample.

Figure 3.5 Wheel Tracking Device for Rutting Susceptibility Test

16
 Chapter 4

RESULTS AND DISCUSSIONS

4.1. INTRODUCTION

The samples were tested for rutting susceptibility. The number of samples was forty
five (45). The binder were neat asphalt, polymer modified bitumen (PMA) and
polypropylene fiber (PP fiber). temperature for testing were 50°C, 55°C & 60°C.for the
statistical model Minitab were used .

4.2. RUT DEPTH (CONSTANT TEMPERATURE& VARIABLE

POLYMERTYPE)

Rut depth is highly dependent on polymer type. The generalize regression equation
above depicts the importance of poly type on rut depth. The study also aims to have a
result of rut depth at constant temperature. The only variation is the polymer types.

Three graph which shows the variation of rut depth on the modification of asphalt at
constant temperature of 50°C, 55°C and 60°C.

Rut Depth - 50°C

6

5.30

5 4.70

4
R.D (mm)

Controlled
2.99
3 PP Fiber
PMA

Figure 4.1 Rut Depths at 50°C

17
 Rut Depth - 55°C

10
9.34
9

7 6.77

6
Controlled
R.D (mm)

5 PP Fiber
4.16 PMA
4

Figure 4.2 Rut Depths at 55°C

Rut Depth - 60°C

18
15.88
16

14

12
10.36
R.D (mm)

10 Controlled
PP Fiber
8 7.11 PMA

Figure 4.3 Rut Depths at 60°C

18
The three of the graph showing the same parameter that Neat asphalt having the greatest
rut depth value as compare to the both PP fiber and PMA at all temperature levels.
PMA has the lowest value of it against all temperature range. As the temperature
increases the rut depth increases in all cases. Effect of temperature is same for all type
of addition in asphalt along with the neat asphalt.

4.3. RUT DEPTH (SAME BINDER TYPE & VARIABLE

TEMPERATURES)

Rut depth is highly dependent on temperature. The generalize regression equation no.
depicts the importance of temperature on rut depth. The study also aims to have a result
of rut depth at variant temperature on different type of asphalt.

Three graph which shows the variation of rut depth on the basis of the temperature
keeping the asphalt type constant.

Rut Depth - Neat Asphalt

18
15.88
16

14

12
R.D (mm)

10 9.34 Controlled samples at 50°C

Controlled samples at 55°C
8 Controlled samples at 60°C

6 5.3

Figure 4.4 Rut depths of Neat Asphalt at variable temperatures

19
 Rut Depth - PP fiber
12

10.36
10 9.34

8
R.D (mm)

PP Fiber at 50°C
6 5.3 PP Fiber at 55°C
PP Fiber at 60°C

Figure 4.5 Rut depths of Polypropylene fiber at variable temperatures

Rut Depth - PMA

8
7.11
7

5
R.D (mm)

4.16 PMA Fiber at 50°C

4 PMA Fiber at 55°C
2.99 PMA Fiber at 60°C
3

Figure 4.6 Rut depths of polymer Modified Asphalt at variable temperatures

20
The three of the graph showing the same parameter that by the increment in temperature
the rut depths increase. Rut depth is least for the least temperature while maximum for
the maximum temperature provided. PP fiber mixture resists the change in temperature
more than the two.

4.4. SUMMARY OF GRAPHS

The rutting tests were carried out at three test temperatures of 50°C, 55°C and 60°C. As
mentioned earlier fifteen samples of each of neat asphalt, polypropylene fiber and
Polymer modified asphalt. At each temperature level five samples were tested. Figure 4
.8 gives the values of rut depths at different temperatures and with variable binders.

It is evident from fig 4 .7 8 that the performance of PMA is better among all against
rutting at all temperature selected for the study and 10,000 wheel passes in wheel
tracking Device. The fact is same that the increase in temperature will increase the
rutti
Rut Depths - Summary Graph ng
18 pot
15.88 enti
16
al.
14 Asphalt Institute
Controlled samples at 50°C
limitation for
rutting is 10 mm PP Fiber at 50°C
12
10.36 PMA Fiber at 50°C
R.D (mm)

10 9.34 Controlled sample at 55°C

PP Fiber samples at 55°C
8 7.11 PMA samples at 55°C
6.77
Controlled sample at 60°C
6 5.3
4.70 PP Fiber samples at 60°C
4.16
4 PMA samples at 60°C
2.99

Figure 4.7 Summary Graph

21
 Table 4.1 Rutting Results

Rut Depth, mm
Temperature,º C
Control PP PMA
50 4.8 4.4 2.7
50 5.0 4.5 3.3
50 5.3 4.7 3.0
50 5.5 4.8 2.5
50 5.8 5.0 3.3
55 9.0 6.4 4.5
55 8.7 6.6 3.5
55 9.3 6.7 4.3
55 9.6 6.9 4.3
55 9.8 7.1 4.0
60 15.5 10.0 7.3
60 15.2 10.1 7.6
60 15.8 10.3 6.8
60 16.1 10.5 6.6
60 16.4 10.7 7.1

22
 Chapter 5

MODEL PREPARATION AND VALIDATION

5.1. MODEL PREPARATION

Model prepare from the result obtained in experiments is one the core work in this
study. Empirical modeling is done using analysis done on Minitab software. Variable
involved in the empirical modeling are temperature and polymer type. Thirty two
different combinations of variables are used starting from only constant to quadratic
form of both variables with the incorporation of multiple terms of both variables
discussed above.

The list of equations is available in annexure. When the model validation curve passes
through origin, the study restricted these combinations. It also has five different variable
in empirical modeling i.e. temperature (T), polymer type (PT), square of temperature
(T2), square of polymer type (PT2) and multiple term of polymer type and temperature
(PT x T). By rules of permutation and combination 25= 32.

5.2. CODING FOR EMPIRICAL MODELING

The following experimental design was adopted for empirical modeling. That includes
 Polymer Type-3 levels
i.e. No polymer, Polypropylene fibers, SBS polymer
 Temperature-3 levels
i.e. 50°C, 55°C, 60°C.
 Therefore total number of experiment with 5 replicates is given as 3x3x5=45
samples
The code adopted for calculation is as follows

Table 5.1 Notation for the Regression Equation

Variable Symbol Actual Values Coded values
Temperature A 50, 55 and 60 +1,0,-1
Polymer Addition B No Polymer, PP, PMA +1,0,-1

23
5.3. DESIGN OF EXPERIMENT MATRIX

The design matrix is given as

Table 5.2 Design of Experiment
Run Order A B Run Order A B
1 1 1 23 0 0
2 1 1 24 0 0
3 1 1 25 0 0
4 1 1 26 0 -1
5 1 1 27 0 -1
6 1 0 28 0 -1
7 1 0 29 0 -1
8 1 0 30 0 -1
9 1 0 31 -1 1
10 1 0 32 -1 1
11 1 -1 33 -1 1
12 1 -1 34 -1 1
13 1 -1 35 -1 1
14 1 -1 36 -1 0
15 1 -1 37 -1 0
16 0 1 38 -1 0
17 0 1 39 -1 0
18 0 1 40 -1 0
19 0 1 41 -1 -1
20 0 1 42 -1 -1
21 0 0 43 -1 -1
22 0 0 44 -1 -1
45 -1 -1

The Results when analyzed for rut depth in Minitab. The elaboration of design of
experiment is for say it is run order number 22 with the value in column A is “0” and
the value in column B is also “0”.The zero in Colum A stands for temperature 55°C and

24
the zero in column B stands for polypropylene. Likewise all values in column A stand
for temperature and for the next column the denotation stands for Asphalt type.

5.4. REGRESSION EQUATION SELECTION AND VALIDATION

The Null Hypothesis of the study is

 Polymer and temperature do affect the rut depth.

The equation having the highest value of R2 for both regression and model validation is
selected as an empirical equation of the study. The equation that qualifies the criteria is
written bellow as the model equation of the study. All the other thirty one equations
along with their ANOVA results are presented in annexure while the ANOVA result of
the model is also written below.

Where,

R.D = Rut Depth in (mm)

T = Temperature (°C)

PT = Polymer Type.

5.5. ANALYSIS OF VARIANCE

Table 5.3 Test Results of ANOVA

Source DF SS SS F P
Regression 5 627.11 125.42 489.26 0.000
Residual 39 10.00 0.26
Error
Lack of Fit 3 5.18 1.73 12.92 0.000
Pure Error 36 4.82 0.13
Total 44 637.11

The above equation is the final regression equation form the data set observed in
laboratory experiments conducted. P value of the regression equation is 0.0 which is
quite satisfactory along with the R square value that 98.4.0 % which is again a

25
satisfactory one. P value = 0.000, Since the P value is very low < 0.005. So, the null
hypothesis is accepted.

Model Validation Curve

18
Avg. Rut value from predicted model

16 R² = 0.98

14

12

10

0
0 5 10 15 20

Avg. Experimental rut values

Figure 5.1 Model Validation Curve

The validation curve passes through origin and having the R2 value 98.4 %. With all
respect the P value, R2 value of the equation and R2 value of model validation curve, the
above mentioned equation is the best description of the study.

26
 Chapter 6

CONCLUSIONS & RECOMMENDATIONS

6.1. CONCLUSIONS

The results of the experimental work and analysis carried out in this study will be
discussed in this section. The following conclusions can be made based on the results of
this research:

 The rut depth is found to increase with increase in test temperature for the
neat asphalt samples. Also, the behavior is same for modified asphalt.
 The differential rate of increments is lesser in case of modified asphalt
binders.
 Comparing Polypropylene fiber performed with Polymer modified asphalt,
the performance of PMA is quite satisfactory.
 PMA mix has rutting about half of the neat asphalt mix and about one third
of the PP fiber one at all selected temperatures.

6.2. RECOMMENDATIONS

Based on the scope of work undertaken in this study, following recommendations are
being made in order to evaluate complete performance modifications in asphalt:

1) Other lengths of fibers are used in PP mixture, other lights should be

incorporated and their effects should be studied.
2) One type of modifier i.e. SBS is used in PMA other more modifiers can be
incorporated for the future development.
3) This study has focused on resisting rutting in asphalt concrete. Other distresses
such as fatigue cracking should also be investigated for samples.
4) Remarkable laboratory testing results demand that modified asphalt should be
tested in test-section of pavement to determine its actual field performance.

27
 REFERENCES

1. Javed F. “Sustainable Financing for the Maintenance of Pakistan’s

Highways”. Journal of Transport and Communications Bulletin for Asia and the
Pacific, 2005;75: p. 83-96.

2. Javed F. “Sustainable Financing for the Maintenance of Pakistan’s

Highways”. Journal of Transport and Communications Bulletin for Asia and the
Pacific, 2005;75: p. 83-96.

3. Pakistan Bureau of Statistics, Pakistan Economic Survey 2010-11,

http://www.scribd.com/doc/56933452/Pakistan-Economic-Survey-2010-2011 last
visited 23/7/12.

4. Airey GD. “Fundamental binder and practical mixture evaluation of

polymer modified bituminous”. The International Journal of Pavement Engineering,
2004;5(3):137-151

5. Uzan J. “Permanent Deformation in Flexible Pavements”. Journal of

Transportation Engineering 2004; 130(1): p. 6−13.

6. Brown SF. Rowlett RD. Boucher JL. “Asphalt modification. In:

Proceedings of Conference on U.S. SHRP Highway Research Program”: Sharing
the Benefits. 1990. p. 181–203.

7. Jew P. Woodhams RT. Polyethylene modified asphalt cement for paving

applications.In: Proceedings of Asphalt Paving Technologists. St. Paul, Minnesota.
1986; p. 541-562.

8. Lu X. Isacsson U. Chemical and Rheological Characteristics of Styrene-

Butadiene-Styrene Polymer-Modified Bitumen.Transportation Research Board.
1999: p. 83-92.

9. Glover JC. Davison R.R. DomkeHC.Ruan Y. Juristyarini P. Knorr BD.

Jung HS.“Development of a new method for assessing asphalt binder durability with
field validation”. Report FHWA/TX-05/1872-2; 2005. Texas Transportation
Institute, The Texas A&M University System, College Station, Texas.

28
10. Ebrahimi GM. “The Effect of Polypropylene Modification on Marshall
Stability and Flow”.MastersThesis.Eastern Mediterranean University, North
Cyprus, 2010.

11. Tapkin S. The effect of Polypropylene fibers on asphalt performance.

Journal of Building and Environment 2007;43(2008): p. 1065-1071.

12. Kaloush KE. Zeiada WA. Biligiri KP. Rodezno MC. Reed JX. Evaluation
of Fiber-Reinforced Asphalt Mixtures Using Advanced Material Characterization
Tests, Journal of Testing and Evaluation, 2010;38(4): p. 12.

13. Simpson AL. Kamyar CM. Case study of modified bituminous mixtures:
Somerset, Kentucky. In: Proceeding from Third Materials Engineering Conference.
New York: 1994. p. 88–96.

14. Abtahi MS, Ebrahimi GM, Kunt MM, Hejazi MS, Esfandiarpour S.
“Production of Polypropylene-reinforced Asphalt Concrete Mixtures Based on Dry
Procedure and Superpave Gyratory Compactor” Iranian Polymer Journal
2011;20(10): 813-823.

29
 ANNEXURE

Regression Analysis: Rut versus PT, T

The regression equation is

Rut00 = 7.40 + 2.71 PT - 3.39

Predictor Coef SE Coef T P

Constant 7.4045 0.1951 37.95 0.000
PT 2.7077 0.2389 11.33 0.000
T -3.3923 0.2389 -14.20 0.000
S = 1.30874 R-Sq = 88.7% R-Sq(adj) = 88.2%

Analysis of Variance

Source DF SS MS F P
Regression 2 565.17 282.58 164.98 0.000
Residual Error 42 71.94 1.71
Lack of Fit 6 67.12 11.19 83.64 0.000
Pure Error 36 4.82 0.13
Total 44 637.11
Source DF Seq SS
PT 1 219.94
T 1 345.22

Unusual Observations

Obs PT Rut Fit SE Fit Residual St Resid

34 1.00 16.180 13.504 0.390 2.676 2.14R
35 1.00 16.480 13.504 0.390 2.976 2.38R

30
Regression Analysis: Rut versus PT, T^2

The regression equation is

Rut01 = 6.76 + 2.71 PT + 0.968 T^2

Predictor Coef SE Coef T P

Constant 6.7592 0.8045 8.40 0.000
PT 2.7077 0.5689 4.76 0.000
T^2 0.9678 0.9854 0.98 0.332
S = 3.11599 R-Sq = 36.0% R-Sq(adj) = 32.9%

Analysis of Variance

Source DF SS MS F P
Regression 2 229.31 114.66 11.81 0.000
Residual Error 42 407.79 9.71
Lack of Fit 3 0.76 0.25 0.02 0.995
Pure Error 39 407.03 10.44
Total 44 637.11
Source DF Seq SS
PT 1 219.94
T^2 1 9.37

Unusual Observations

Obs PT Rut01 Fit SE Fit Residual St Resid

35 1.00 16.480 10.435 0.805 6.045 2.01R

31
Regression Analysis: Rut versus PT ^2, T

The regression equation is

Rut02 = 7.28 + 0.184 PT ^2 - 3.39 T

Predictor Coef SE Coef T P

Constant 7.2820 0.6803 10.70 0.000
PT ^2 0.1836 0.8332 0.22 0.827
T -3.3923 0.4810 -7.05 0.000
S = 2.63468 R-Sq = 54.2% R-Sq(adj) = 52.1%

Analysis of Variance

Source DF SS MS F P
Regression 2 345.56 172.78 24.89 0.000
Residual Error 42 291.54 6.94
Lack of Fit 3 14.33 4.78 0.67 0.574
Pure Error 39 277.21 7.11
Total 44 637.11
Source DF Seq SS
PT ^2 1 0.34
T 1 345.22

Unusual Observations

Obs PT ^2 Rut02 Fit SE Fit Residual St Resid

34 1.00 16.180 10.858 0.680 5.322 2.09R
35 1.00 16.480 10.858 0.680 5.622 2.21R

32
Regression Analysis: Rut versus PT ^2, T^2

The regression equation is

Rut03 = 6.64 + 0.18 PT ^2 + 0.97 T^2

Predictor Coef SE Coef T P

Constant 6.637 1.288 5.15 0.000
PT ^2 0.184 1.222 0.15 0.881
T^2 0.968 1.222 0.79 0.433
S = 3.86499 R-Sq = 1.5% R-Sq(adj) = 0.0%

Analysis of Variance

Source DF SS MS F P
Regression 2 9.70 4.85 0.32 0.724
Residual Error 42 627.40 14.94
Lack of Fit 1 0.21 0.21 0.01 0.908
Pure Error 41 627.20 15.30
Total 44 637.11
Source DF Seq SS
PT ^2 1 0.34
T^2 1 9.3

Unusual Observations

Obs PT ^2 Rut03 Fit SE Fit Residual St Resid

31 1.00 15.580 7.788 0.815 7.792 2.06R
33 1.00 15.880 7.788 0.815 8.092 2.14R
34 1.00 16.180 7.788 0.815 8.392 2.22R
35 1.00 16.480 7.788 0.815 8.692 2.30R

33
Regression Analysis: Rut versus T

The regression equation is

Rut04 = 7.40 - 3.39 T

Predictor Coef SE Coef T P

Constant 7.4045 0.3884 19.06 0.000
T -3.3923 0.4757 -7.13 0.000
S = 2.60537 R-Sq = 54.2% R-Sq(adj) = 53.1%

Analysis of Variance

Source DF SS MS F P
Regression 1 345.22 345.22 50.86 0.000
Residual Error 43 291.88 6.79
Lack of Fit 1 9.37 9.37 1.39 0.245
Pure Error 42 282.51 6.73
Total 44 637.11

Unusual Observations

Obs T Rut04 Fit SE Fit Residual St Resid

33 -1.00 15.880 10.797 0.614 5.083 2.01R
34 -1.00 16.180 10.797 0.614 5.383 2.13R
35 -1.00 16.480 10.797 0.614 5.683 2.24R

34
Regression Analysis: Rut versus PT

The regression equation is

Rut05 = 7.40 + 2.71 PT

Predictor Coef SE Coef T P

Constant 7.4045 0.4643 15.95 0.000
PT 2.7077 0.5687 4.76 0.000
S = 3.11471 R-Sq = 34.5% R-Sq(adj) = 33.0%

Analysis of Variance

Source DF SS MS F P
Regression 1 219.94 219.94 22.67 0.000
Residual Error 43 417.16 9.70
Lack of Fit 1 0.34 0.34 0.03 0.855
Pure Error 42 416.82 9.92
Total 44 637.11

Unusual Observations

Obs PT Rut05 Fit SE Fit Residual St Resid

34 1.00 16.180 10.112 0.734 6.068 2.00R
35 1.00 16.480 10.112 0.734 6.368 2.10R

35
Regression Analysis: Rut versus T^2, T, PT

The regression equation is

Rut06 = 6.76 + 0.968 T^2 - 3.39 T + 2.71 PT

Predictor Coef SE Coef T P

Constant 6.7592 0.3190 21.19 0.000
T^2 0.9678 0.3907 2.48 0.017
T -3.3923 0.2255 -15.04 0.000
PT 2.7077 0.2255 12.01 0.000
S = 1.23536 R-Sq = 90.2% R-Sq(adj) = 89.5%

Analysis of Variance

Source DF SS MS F P
Regression 3 574.53 191.51 125.49 0.000
Residual Error 41 62.57 1.53
Lack of Fit 5 57.76 11.55 86.36 0.000
Pure Error 36 4.82 0.13
Total 44 637.11
Source DF Seq SS
T^2 1 9.37
T 1 345.22
PT 1 219.94

Unusual Observations

Obs T^2 Rut06 Fit SE Fit Residual St Resid

34 1.00 16.180 13.827 0.391 2.353 2.01R
35 1.00 16.480 13.827 0.391 2.653 2.26R

36
Regression Analysis: Rut versus PT ^2, PT, T

The regression equation is

Rut07 = 7.28 + 0.184 PT ^2 + 2.71 PT - 3.39 T

Predictor Coef SE Coef T P

Constant 7.2820 0.3412 21.34 0.000
PT ^2 0.1836 0.4179 0.44 0.663
PT 2.7077 0.2413 11.22 0.000
T -3.3923 0.2413 -14.06 0.000
S = 1.32149 R-Sq = 88.8% R-Sq(adj) = 87.9%

Analysis of Variance

Source DF SS MS F P
Regression 3 565.51 188.50 107.94 0.000
Residual Error 41 71.60 1.75
Lack of Fit 5 66.79 13.36 99.87 0.000
Pure Error 36 4.82 0.13
Total 44 637.11
Source DF Seq SS
PT ^2 1 0.34
PT 1 219.94
T 1 345.22

Unusual Observations

Obs PT ^2 Rut07 Fit SE Fit Residual St Resid

34 1.00 16.180 13.566 0.418 2.614 2.09R
35 1.00 16.480 13.566 0.418 2.914 2.32R

37
Regression Analysis: Rut versus PT ^2, T^2, PT, T

The regression equation is

Rut08 = 6.64 + 0.184 PT ^2 + 0.968 T^2 + 2.71 PT - 3.39 T

Predictor Coef SE Coef T P

Constant 6.6368 0.4158 15.96 0.000
PT ^2 0.1836 0.3944 0.47 0.644
T^2 0.9678 0.3944 2.45 0.019
PT 2.7077 0.2277 11.89 0.000
T -3.3923 0.2277 -14.90 0.000
S = 1.24733 R-Sq = 90.2% R-Sq(adj) = 89.3%

Analysis of Variance

Source DF SS MS F P
Regression 4 574.87 143.72 92.37 0.000
Residual Error 40 62.23 1.56
Lack of Fit 4 57.42 14.35 107.32 0.000
Pure Error 36 4.82 0.13
Total 44 637.11
Source DF Seq SS
PT ^2 1 0.34
T^2 1 9.37
PT 1 219.94
T 1 345.22

Unusual Observations

Obs PT ^2 Rut08 Fit SE Fit Residual St Resid

35 1.00 16.480 13.888 0.416 2.592 2.20R

38
Regression Analysis: Rut versus PT ^2, PT, T^2

The regression equation is

Rut09 = 6.64 + 0.184 PT ^2 + 2.71 PT + 0.968 T^2

Predictor Coef SE Coef T P

Constant 6.637 1.051 6.32 0.000
PT ^2 0.1836 0.9969 0.18 0.855
PT 2.7077 0.5756 4.70 0.000
T^2 0.9678 0.9969 0.97 0.337
S = 3.15246 R-Sq = 36.0% R-Sq(adj) = 31.4%

Analysis of Variance

Source DF SS MS F P
Regression 3 229.648 76.549 7.70 0.000
Residual Error 41 407.457 9.938
Lack of Fit 2 0.424 0.212 0.02 0.980
Pure Error 39 407.034 10.437
Total 44 637.105

Source DF Seq SS
PT ^2 1 0.337
PT 1 219.944
T^2 1 9.367
Regression Analysis: Rut versus PT ^2, T^2, T

The regression equation is

Rut10 = 6.64 + 0.184 PT ^2 + 0.968 T^2 - 3.39 T

Predictor Coef SE Coef T P

Constant 6.6368 0.8745 7.59 0.000
PT ^2 0.1836 0.8296 0.22 0.826
39
T^2 0.9678 0.8296 1.17 0.250
T -3.3923 0.4790 -7.08 0.000
S = 2.62343 R-Sq = 55.7% R-Sq(adj) = 52.5

Analysis of Variance

Source DF SS MS F P
Regression 3 354.93 118.31 17.19 0.000
Residual Error 41 282.18 6.88
Lack of Fit 2 4.96 2.48 0.35 0.707
Pure Error 39 277.21 7.11
Total 44 637.11
Source DF Seq SS
PT ^2 1 0.34
T^2 1 9.37
T 1 345.22

Unusual Observations

Obs PT ^2 Rut10 Fit SE Fit Residual St Resid

35 1.00 16.480 11.181 0.732 5.299 2.10R

Regression Analysis: RUT versus PT ^2

The regression equation is

RUT11 = 7.28 + 0.18 PT ^2

Predictor Coef SE Coef T P

Constant 7.2820 0.9936 7.33 0.000
PT ^2 0.184 1.217 0.15 0.881
S = 3.84819 R-Sq = 0.1% R-Sq(adj) = 0.0%

40
Analysis of Variance

Source DF SS MS F P
Regression 1 0.34 0.34 0.02 0.881
Residual Error 43 636.77 14.81
Total 44 637.11
The number of distinct predictor combinations equals the number of parameters.
No degrees of freedom for lack of fit.

Unusual Observations

Obs PT ^2 RUT11 Fit SE Fit Residual St Resid

31 1.00 15.580 7.466 0.703 8.114 2.14R
32 1.00 15.280 7.466 0.703 7.814 2.07R
33 1.00 15.880 7.466 0.703 8.414 2.22R
34 1.00 16.180 7.466 0.703 8.714 2.30R
35 1.00 16.480 7.466 0.703 9.014 2.38R

Regression Analysis: Rut versus PT ^2

The regression equation is

Rut12 = 7.28 + 0.18 PT ^2

Predictor Coef SE Coef T P

Constant 7.2820 0.9936 7.33 0.000
PT ^2 0.184 1.217 0.15 0.881
S = 3.84819 R-Sq = 0.1% R-Sq(adj) = 0.0%

Analysis of Variance

Source DF SS MS F P
Regression 1 0.34 0.34 0.02 0.881
Residual Error 43 636.77 14.81
Total 44 637.11

41
The number of distinct predictor combinations equals the number of parameters.
No degrees of freedom for lack of fit.
Cannot do pure error test.

Unusual Observations

Obs PT ^2 Rut12 Fit SE Fit Residual St Resid

31 1.00 15.580 7.466 0.703 8.114 2.14R
32 1.00 15.280 7.466 0.703 7.814 2.07R
33 1.00 15.880 7.466 0.703 8.414 2.22R
34 1.00 16.180 7.466 0.703 8.714 2.30R
35 1.00 16.480 7.466 0.703 9.014 2.38R

Regression Analysis: Rut versus T^2

The regression equation is

Rut12 = 6.76 + 0.97 T^2

Predictor Coef SE Coef T P

Constant 6.7592 0.9865 6.85 0.000
T^2 0.968 1.208 0.80 0.428
S = 3.82081 R-Sq = 1.5% R-Sq(adj) = 0.0%

Analysis of Variance

Source DF SS MS F P
Regression 1 9.37 9.37 0.64 0.428
Residual Error 43 627.74 14.60
Total 44 637.11

The number of distinct predictor combinations equals the number of parameters.

No degrees of freedom for lack of fit.
Cannot do pure error test.

Unusual Observations
42
Obs T^2 Rut12 Fit SE Fit Residual St Resid
31 1.00 15.580 7.727 0.698 7.853 2.09R
32 1.00 15.280 7.727 0.698 7.553 2.01R
33 1.00 15.880 7.727 0.698 8.153 2.17R
34 1.00 16.180 7.727 0.698 8.453 2.25R
35 1.00 16.480 7.727 0.698 8.753 2.33R

Regression Analysis: Rut versus PT ^2, PT

The regression equation is

Rut13 = 7.28 + 0.184 PT ^2 + 2.71 PT

Predictor Coef SE Coef T P

Constant 7.2820 0.8134 8.95 0.000
PT ^2 0.1836 0.9962 0.18 0.855
PT 2.7077 0.5752 4.71 0.000
S = 3.15030 R-Sq = 34.6% R-Sq(adj) = 31.5%

Analysis of Variance

Source DF SS MS F P
Regression 2 220.28 110.14 11.10 0.000
Residual Error 42 416.82 9.92
Total 44 637.11

The number of distinct predictor combinations equals the number of parameters.

No degrees of freedom for lack of fit.
Cannot do pure error test.
Source DF Seq SS
PT ^2 1 0.34
PT 1 219.94

43
Unusual Observations

Obs PT ^2 Rut13 Fit SE Fit Residual St Resid

35 1.00 16.480 10.173 0.813 6.307 2.07R

Regression Analysis: Rut versus T^2, T

The regression equation is

Rut14 = 6.76 + 0.968 T^2 - 3.39 T

Predictor Coef SE Coef T P

Constant 6.7592 0.6697 10.09 0.000
T^2 0.9678 0.8202 1.18 0.245
T -3.3923 0.4735 -7.16 0.000
S = 2.59356 R-Sq = 55.7% R-Sq(adj) = 53.5%

Analysis of Variance

Source DF SS MS F P
Regression 2 354.59 177.30 26.36 0.000
Residual Error 42 282.51 6.73
Total 44 637.11

The number of distinct predictor combinations equals the number of parameters.

No degrees of freedom for lack of fit.
Cannot do pure error test.
Source DF Seq SS
T^2 1 9.37
T 1 345.22

Unusual Observations

Obs T^2 Rut14 Fit SE Fit Residual St Resid

34 1.00 16.180 11.119 0.670 5.061 2.02R
35 1.00 16.480 11.119 0.670 5.361 2.14
44
Regression Analysis: Rut versus PT ^2, PT, T^2, T, T x PT

The regression equation is

Rut15 = 6.64 + 0.184 PT ^2 + 2.71 PT + 0.968 T^2 - 3.39 T - 1.62 T x PT

Predictor Coef SE Coef T P

Constant 6.6368 0.1688 39.32 0.000
PT ^2 0.1836 0.1601 1.15 0.258
PT 2.70767 0.09244 29.29 0.000
T^2 0.9678 0.1601 6.04 0.000
T -3.39227 0.09244 -36.70 0.000
T x PT -1.6161 0.1132 -14.27 0.000
S = 0.506310 R-Sq = 98.4% R-Sq(adj) = 98.2%

Analysis of Variance

Source DF SS MS F P
Regression 5 627.11 125.42 489.26 0.000
Residual Error 39 10.00 0.26
Lack of Fit 3 5.18 1.73 12.92 0.000
Pure Error 36 4.82 0.13
Total 44 637.11
Source DF Seq SS
PT ^2 1 0.34
PT 1 219.94
T^2 1 9.37
T 1 345.22
T x PT 1 52.24

Unusual Observations

Obs PT ^2 Rut15 Fit SE Fit Residual St Resid

35 1.00 16.4800 15.5043 0.2032 0.9757 2.10R
36 0.00 10.0000 10.9969 0.1688 -0.9969 -2.09R

45
Regression Analysis: Rut versus T^2, T, PT, T x PT

The regression equation is

Rut16 = 6.76 + 0.968 T^2 - 3.39 T + 2.71 PT - 1.62 T x PT

Predictor Coef SE Coef T P

Constant 6.7592 0.1312 51.50 0.000
T^2 0.9678 0.1607 6.02 0.000
T -3.39227 0.09280 -36.55 0.000
PT 2.70767 0.09280 29.18 0.000
T x PT -1.6161 0.1137 -14.22 0.000
S = 0.508303 R-Sq = 98.4% R-Sq(adj) = 98.2%

Analysis of Variance

Source DF SS MS F P
Regression 4 626.77 156.69 606.46 0.000
Residual Error 40 10.33 0.26
Lack of Fit 4 5.52 1.38 10.32 0.000
Pure Error 36 4.82 0.13
Total 44 637.11
Source DF Seq SS
T^2 1 9.37
T 1 345.22
PT 1 219.94
T x PT 1 52.24

Unusual Observations

Obs T^2 Rut16 Fit SE Fit Residual St Resid

35 1.00 16.4800 15.4431 0.1969 1.0369 2.21R
36 1.00 10.0000 11.1193 0.1312 -1.1193 -2.28R

R denotes an observation with a large standardized residual

46
Regression Analysis: Rut versus PT, T, T x PT

The regression equation is

Rut17 = 7.40 + 2.71 PT - 3.39 T - 1.62 T x PT

Predictor Coef SE Coef T P

Constant 7.4045 0.1033 71.65 0.000
PT 2.7077 0.1266 21.39 0.000
T -3.3923 0.1266 -26.80 0.000
T x PT -1.6161 0.1550 -10.43 0.000
S = 0.693205 R-Sq = 96.9% R-Sq(adj) = 96.7%

Analysis of Variance

Source DF SS MS F P
Regression 3 617.40 205.80 428.28 0.000
Residual Error 41 19.70 0.48
Lack of Fit 5 14.89 2.98 22.26 0.000
Pure Error 36 4.82 0.13
Total 44 637.1
Source DF Seq SS
PT 1 219.94
T 1 345.22
T x PT 1 52.24

Unusual Observations

Obs PT Rut17 Fit SE Fit Residual St Resid

35 1.00 16.480 15.120 0.258 1.360 2.11R

R denotes an observation with a large standardized residual.

47
Regression Analysis: Rut versus PT, T^2, T x PT

The regression equation is

Rut18 = 6.76 + 2.71 PT + 0.968 T^2 - 1.62 T x PT

Predictor Coef SE Coef T P

Constant 6.7592 0.7604 8.89 0.000
PT 2.7077 0.5377 5.04 0.000
T^2 0.9678 0.9312 1.04 0.305
T x PT -1.6161 0.6585 -2.45 0.018
S = 2.94486 R-Sq = 44.2% R-Sq(adj) = 40.1%

Analysis of Variance

Source DF SS MS F P
Regression 3 281.546 93.849 10.82 0.000
Residual Error 41 355.559 8.672
Lack of Fit 4 270.712 67.678 29.51 0.000
Pure Error 37 84.847 2.293
Total 44 637.105
Source DF Seq SS
PT 1 219.944
T^2 1 9.367
T x PT 1 52.236

48
Regression Analysis: Rut versus PT ^2, T, T x PT

The regression equation is

Rut19 = 7.28 + 0.184 PT ^2 - 3.39 T - 1.62 T x PT

Predictor Coef SE Coef T P

Constant 7.2820 0.6238 11.67 0.000
PT ^2 0.1836 0.7640 0.24 0.811
T -3.3923 0.4411 -7.69 0.000
T x PT -1.6161 0.5402 -2.99 0.005
S = 2.41595 R-Sq = 62.4% R-Sq(adj) = 59.7%

Analysis of Variance

Source DF SS MS F P
Regression 3 397.80 132.60 22.72 0.000
Residual Error 41 239.31 5.84
Lack of Fit 4 167.56 41.89 21.60 0.000
Pure Error 37 71.75 1.94
Total 44 637.11
Source DF Seq SS
PT ^2 1 0.34
T 1 345.22
T x PT 1 52.24

Regression Analysis: Rut versus T^2, PT ^2, T x PT

The regression equation is

Rut20 = 6.64 + 0.97 T^2 + 0.18 PT ^2 - 1.62 T x PT

Predictor Coef SE Coef T P

Constant 6.637 1.248 5.32 0.000
T^2 0.968 1.184 0.82 0.419
PT ^2 0.184 1.184 0.16 0.878

49
T x PT -1.6161 0.8375 -1.93 0.061
S = 3.74545 R-Sq = 9.7% R-Sq(adj) = 3.1%

Analysis of Variance

Source DF SS MS F P
Regression 3 61.94 20.65 1.47 0.236
Residual Error 41 575.17 14.03
Lack of Fit 1 0.21 0.21 0.01 0.905
Pure Error 40 574.96 14.37
Total 44 637.11
Source DF Seq SS
T^2 1 9.37
PT ^2 1 0.34
T x PT 1 52.24

Regression Analysis: Rut versus T, T x PT

The regression equation is

Rut21 = 7.40 - 3.39 T - 1.62 T x PT

Predictor Coef SE Coef T P

Constant 7.4045 0.3561 20.79 0.000
T -3.3923 0.4361 -7.78 0.000
T x PT -1.6161 0.5341 -3.03 0.00
S = 2.38869 R-Sq = 62.4% R-Sq(adj) = 60.6%

Analysis of Variance

Source DF SS MS F P
Regression 2 397.46 198.73 34.83 0.000
Residual Error 42 239.65 5.71
Lack of Fit 4 167.89 41.97 22.23 0.000
Pure Error 38 71.75 1.89
Total 44 637.11
50
Source DF Seq SS
T 1 345.22
T x PT 1 52.24

Regression Analysis: Rut22 versus PT, T x PT

The regression equation is

Rut22 = 7.40 + 2.71 PT - 1.62 T x PT

Predictor Coef SE Coef T P

Constant 7.4045 0.4394 16.85 0.000
PT 2.7077 0.5382 5.03 0.000
T x PT -1.6161 0.6591 -2.45 0.018
S = 2.94766 R-Sq = 42.7% R-Sq(adj) = 40.0%

Analysis of Variance

Source DF SS MS F P
Regression 2 272.18 136.09 15.66 0.000
Residual Error 42 364.93 8.69
Lack of Fit 4 278.13 69.53 30.44 0.000
Pure Error 38 86.80 2.28
Total 44 637.11
Source DF Seq SS
PT 1 219.94
T x PT 1 52.24

Regression Analysis: Rut versus PT ^2, PT, T, T x PT

The regression equation is

Rut23 = 7.28 + 0.184 PT ^2 + 2.71 PT - 3.39 T - 1.62 T x PT

Predictor Coef SE Coef T P

Constant 7.2820 0.1797 40.53 0.000
PT ^2 0.1836 0.2200 0.83 0.409
51
PT 2.7077 0.1270 21.31 0.000
T -3.3923 0.1270 -26.70 0.000
T x PT -1.6161 0.1556 -10.39 0.000
S = 0.695785 R-Sq = 97.0% R-Sq(adj) = 96.7%

Analysis of Variance

Source DF SS MS F P
Regression 4 617.74 154.44 319.00 0.000
Residual Error 40 19.36 0.48
Lack of Fit 4 14.55 3.64 27.20 0.000
Pure Error 36 4.82 0.13
Total 44 637.11

Source DF Seq SS
PT ^2 1 0.34
PT 1 219.94
T 1 345.22
T x PT 1 52.24

Unusual Observations

Obs PT ^2 Rut23 Fit SE Fit Residual St Resid

17 1.00 8.790 10.173 0.180 -1.383 -2.06R
35 1.00 16.480 15.182 0.269 1.298 2.02R

Regression Analysis: Rut versus PT ^2, T^2, PT, T x PT

The regression equation is

Rut24 = 6.64 + 0.184 PT ^2 + 0.968 T^2 + 2.71 PT - 1.62 T x PT

Predictor Coef SE Coef T P

Constant 6.6368 0.9933 6.68 0.000
PT ^2 0.1836 0.9424 0.19 0.846
52
T^2 0.9678 0.9424 1.03 0.311
PT 2.7077 0.5441 4.98 0.000
T x PT -1.6161 0.6664 -2.43 0.020
S = 2.98002 R-Sq = 44.2% R-Sq(adj) = 38.7%

Analysis of Variance

Source DF SS MS F P
Regression 4 281.884 70.471 7.94 0.000
Residual Error 40 355.222 8.881
Lack of Fit 3 270.374 90.125 39.30 0.000
Pure Error 37 84.847 2.293
Total 44 637.105
Source DF Seq SS
PT ^2 1 0.337
T^2 1 9.367
PT 1 219.944
T x PT 1 52.236

Regression Analysis: Rut versus PT ^2, T^2, T, T x PT

The regression equation is

Rut25 = 6.64 + 0.184 PT ^2 + 0.968 T^2 - 3.39 T - 1.62 T x PT

Predictor Coef SE Coef T P

Constant 6.6368 0.7992 8.30 0.000
PT ^2 0.1836 0.7582 0.24 0.810
T^2 0.9678 0.7582 1.28 0.209
T -3.3923 0.4377 -7.75 0.000
T x PT -1.6161 0.5361 -3.01 0.004
S = 2.39761 R-Sq = 63.9% R-Sq(adj) = 60.3%

Analysis of Variance

53
Source DF SS MS F P
Regression 4 407.16 101.79 17.71 0.000
Residual Error 40 229.94 5.75
Lack of Fit 3 158.19 52.73 27.19 0.000
Pure Error 37 71.75 1.94
Total 44 637.11

Source DF Seq SS
PT ^2 1 0.34
T^2 1 9.37
T 1 345.22
T x PT 1 52.24

Regression Analysis: Rut versus PT ^2, T x PT

The regression equation is

Rut26 = 7.28 + 0.18 PT ^2 - 1.62 T x PT

Predictor Coef SE Coef T P

Constant 7.2820 0.9632 7.56 0.000
PT ^2 0.184 1.180 0.16 0.877
T x PT -1.6161 0.8342 -1.94 0.059
S = 3.73061 R-Sq = 8.3% R-Sq(adj) = 3.9%

Analysis of Variance

Source DF SS MS F P
Regression 2 52.57 26.29 1.89 0.164
Residual Error 42 584.53 13.92
Lack of Fit 1 7.62 7.62 0.54 0.466
Pure Error 41 576.91 14.07
Total 44 637.11
Source DF Seq SS
PT ^2 1 0.34
54
T x PT 1 52.24

Unusual Observations

Obs PT ^2 Rut26 Fit SE Fit Residual St Resid

35 1.00 16.480 9.082 1.077 7.398 2.07R

Regression Analysis: Rut versus T^2, T x PT

The regression equation is

Rut27 = 6.76 + 0.97 T^2 - 1.62 T x PT

Predictor Coef SE Coef T P

Constant 6.7592 0.9558 7.07 0.000
T^2 0.968 1.171 0.83 0.413
T x PT -1.6161 0.8277 -1.95 0.058

S = 3.70168 R-Sq = 9.7% R-Sq(adj) = 5.4%

Analysis of Variance

Source DF SS MS F P
Regression 2 61.60 30.80 2.25 0.118
Residual Error 42 575.50 13.70
Lack of Fit 1 0.54 0.54 0.04 0.845
Pure Error 41 574.96 14.02
Total 44 637.11

Source DF Seq SS
T^2 1 9.37
T x PT 1 52.24
55
Unusual Observations

Obs T^2 Rut27 Fit SE Fit Residual St Resid

35 1.00 16.480 9.343 1.069 7.137 2.01R

R denotes an observation with a large standardized residual.

Regression Analysis: Rut versus PT ^2, PT, T x PT

The regression equation is

Rut28 = 7.28 + 0.184 PT ^2 + 2.71 PT - 1.62 T x PT

Predictor Coef SE Coef T P

Constant 7.2820 0.7700 9.46 0.000
PT ^2 0.1836 0.9430 0.19 0.847
PT 2.7077 0.5444 4.97 0.000
T x PT -1.6161 0.6668 -2.42 0.020
S = 2.98201 R-Sq = 42.8% R-Sq(adj) = 38.6%

Analysis of Variance

Source DF SS MS F P
Regression 3 272.517 90.839 10.22 0.000
Residual Error 41 364.589 8.892
Lack of Fit 3 277.791 92.597 40.54 0.000
Pure Error 38 86.798 2.284
Total 44 637.105
Source DF Seq SS
PT ^2 1 0.337
PT 1 219.944
T x PT 1 52.236

56
Regression Analysis: Rut versus T^2, T, T x PT

The regression equation is

Rut29 = 6.76 + 0.968 T^2 - 3.39 T - 1.62 T x PT

Predictor Coef SE Coef T P

Constant 6.7592 0.6119 11.05 0.000
T^2 0.9678 0.7494 1.29 0.204
T -3.3923 0.4327 -7.84 0.000
T x PT -1.6161 0.5299 -3.05 0.004
S = 2.36993 R-Sq = 63.9% R-Sq(adj) = 61.2%

Analysis of Variance

Source DF SS MS F P
Regression 3 406.83 135.61 24.14 0.000
Residual Error 41 230.28 5.62
Lack of Fit 3 158.53 52.84 27.99 0.000
Pure Error 38 71.75 1.89
Total 44 637.11
Source DF Seq SS
T^2 1 9.37
T 1 345.22
T x PT 1 52.24

Regression Analysis: Rut versus T x PT

The regression equation is

Rut30 = 7.40 - 1.62 T x PT

Predictor Coef SE Coef T P

Constant 7.4045 0.5498 13.47 0.000
T x PT -1.6161 0.8247 -1.96 0.05
S = 3.68804 R-Sq = 8.2% R-Sq(adj) = 6.1%

57
Analysis of Variance

Source DF SS MS F P
Regression 1 52.24 52.24 3.84 0.057
Residual Error 43 584.87 13.60
Lack of Fit 1 6.28 6.28 0.46 0.503
Pure Error 42 578.59 13.78
Total 44 637.11

Unusual Observations

Obs T x PT Rut30 Fit SE Fit Residual St Resid

34 -1.00 16.180 9.021 0.991 7.159 2.02R
35 -1.00 16.480 9.021 0.991 7.459 2.10R

58