Data Analysis for Pavament Maintenance Asset Managmet

8/3/2019 Data Analysis for Pavament Maintenance Asset Managmet

http://slidepdf.com/reader/full/data-analysis-for-pavament-maintenance-asset-managmet 1/16

DATA ANALYSIS FOR PAVEMENT MAINTENANCE ASSET MANAGEMENT

Marios Papayiannis

1

DATA ANALYSIS FOR PAVEMENT

MAINTENANCE ASSET MANAGEMENT

Contents

1 INTRODUCTION .................................................................................................................................. 2

1.1 Road Networks ............................................................................................................................ 2

1.2 Pavement Maintenance .............................................................................................................. 2

1.3 Asset Management ..................................................................................................................... 2

2 DATA AND ANALYSIS .......................................................................................................................... 5

2.1 Profiles ........................................................................................................................................ 5

2.2 Profiles, Data and Profilers .......................................................................................................... 5

2.2.1 Profile Variance .................................................................................................................... 5

2.2.2 International Roughness Index............................................................................................. 7

2.2.3 Inertial Profilers.................................................................................................................... 9

3 ANALYSIS, DIFFERENCES AND COMPARISSON ................................................................................. 10

3.1 Differences Between Analysis Techniques ................................................................................ 10

3.2 Similarities Between Techniques .............................................................................................. 10

4. FURTHER ANALYSIS FOR DERIVATION OF DETERIORATION MODEL ............................................... 12

4.1 Construction of Deterioration Model ........................................................................................ 12

4.2 Effect of Errors on Prediction Models ....................................................................................... 12

4.3 Other Factors Affecting Deterioration Model ........................................................................... 13

4.4 Minimizing Error ....................................................................................................................... 13

5 CONCLUSION.................................................................................................................................... 15

5.1 Discussion ................................................................................................................................. 15

5.2 Proposal .................................................................................................................................... 15




Marios Papayiannis

2

1 INTRODUCTION

1.1 Road Networks

Highway networks comprise the most crucial infrastructure of any country or

state. Rapid growth in developed countries and the continuous need for time

efficiency, regarding most if not all industries has resulted in transportation

through vehicle media being essential for the wellbeing of every industry and

inevitably every nation, meaning that such infrastructure directly affects the

socioeconomic condition and development of every country. Therefore the wide

use of most road networks and highway networks in particular, establishes

safety and ride quality as top priorities.

1.2 Pavement Maintenance

In order to preserve the condition of any network, maintenance must be

undertaken. Such a procedure will require conservation of the quality of ride and

concurrently the safety which a road provides. Due to the immense size of road

infrastructure, the process of servicing requires major sums of money. Hence

there is a great need to optimize these procedures, both in terms of timing and

money, since maintenance becomes inevitable at some point in the life cycle of a

road.

1.3 Asset Management

Asset management is generally a very broad term with different interpretation

and different understanding from different parties. The term itself could refer to

a fixed asset, managed with various techniques, or targeting the right amount of

money according to a single strategy (County Surveyors’ Society, 2004).

Asset management is a procedure whereby a systematic process is applied to

achieve a long term and cost effective solution. Additionally the overall process

has to be strategic and be able to allocate the necessary resources to the right

place at the right time, in order to maximise efficiency. These resources include

strategies in all aspects, engineering, economic, management and business.

Thus it will be possible to achieve continuous improvement framework, by the

utilisation of technological means (County Surveyors’ Society, 2004).

As managing administrations become increasingly more constrained by

parameters which cannot control or foresee the necessity for accuracy becomes




Marios Papayiannis

3

crucial. The continuous limiting of budgets results in a serious shortage both in

personnel and economic resources (County Surveyors’ Society, 2004).

Therefore, it is crucial to develop known technological and analytical techniques

to maximise the efficiency of road maintenance and assure no unnecessary work

is done without jeopardizing the safety of the roads.




Marios Papayiannis

4




Marios Papayiannis

5

2 DATA AND ANALYSIS

2.1 Profiles

A profile is a section of the road surface, which stretches along a line and can be

both longitudinal and lateral. Both types of profiles can be useful for assessing

the condition and quality of the road surface and roughness (Sayers,

Karamichas, 1998).

Fig 1. Road Profiles (Sayers, Karamichas, 1998)

Longitudinal profiles, illustrate the deviation of the surface of the pavement

along the distance it stretches. It is difficult to repeat the same profile twice,

although it is highly dependent on the size of the profiler instrument. In order for

the most representative profile to be produced therefore, and counteract for the

multiple profiles possible on a single stretch of road, measurements are usually

recorded along wheel track positions, although accuracy can be further increased

by increasing profile lines (Sayers, Karamichas, 1998).

2.2 Profiles, Data and Profilers

2.2.1 Roughness Variance

Profiles are represented graphically in terms of an indicator for the condition and

surface smoothness of the road. One of the main variants used is Roughness

Variance (RV). Following the recording of the longitudinal profile data, by means

of a profiler (e.g SCANNER), it is essential to analyse the data, and investigate

the wavelength content of the profile. By processing these data further, using






Marios Papayiannis

7

This method is highly influenced by geometrical factors, especially as the

variance length is increased. In order to reduce this there has been an effort to

include filters into the whole procedure, to produce the enhanced variance which

limits the influence of geometry and similar parameters onto the output. This

technique has, since 2004, been established as the standard technique in the UK

(Benbow et. al, 2006).

When graphically representing these data, it is suggested that, threshold levels

are assessed and set by the corresponding authority, although there are

indicated levels in the Highways Agency Interim Advice Note 42/02. All values

below the lower threshold are automatically changed to 0 and those above the

higher threshold are given the value of 100. The values which lie in between are

scaled using a linear factor (Benbow et. al, 2006).

2.2.2 International Roughness Index

A more internationally used approach of investigating road condition is by means

of the International Roughness Index or IRI. In the 1970’s the World Bank

financed the development of IRI’s in order to assess whether developing

countries should spend large sums of money in the construction of new

infrastructure, or whether it was possible to sustain the road networks they hadand save assets for other development (Sayers, Karamichas, 1998).

Following this development it was then realised that, due to the differences in

techniques, equipment, weather conditions and other factors, it was impossible

at the time to compare roughness data from different countries. As a result the

World Bank set out an investigation in 1982 to make a calibration and

correlation for these measurements. This gave rise to the measuring scale of

in/mi, proven to be the most suitable for all measurement instruments (Sayers,Karamichas, 1998).

After rigorous research mathematical implementations and computational

methods have been introduced to the technique of calculating IRI’s. Perhaps the

most fundamental part of these techniques is the use of a quarter car model,

which represents the response of the suspension of one wheel of a vehicle. This

allows the representation of the true profile and the vehicle response according

to surface roughness, and additionally is an indicator of vehicle running cost on




Marios Papayiannis

8

that specific road, ride quality, surface condition and dynamic loading effects

from heavy vehicles and sudden braking (Sayers, Karamichas, 1998).

As can be seen from Fig 4 as the road investigated becomes rougher and has

much more surface deviations the value of the IRI increases. Essentially a value

of 0m/km would describe the perfect road and a value above 8m/km represents

a pavement accessible only at low speeds (Sayers, Karamichas, 1998).

Fig 5 IRI values for a specific road (Benbow et. al. 2006)

Fig 3. Quarter car model

(Shahin, 2005)

Fig 4. Different values of IRI’s for

different types of road

(Sayers, Karamichas, 1998)




Marios Papayiannis

9

2.2.3 Inertial Profilers

A particularly important technique of assessing the quality of the road

roughness, apart from the use of laser or ultrasonic SCANNER, is the use of an

inertial profiler. This technique allows for the investigation to be carried out at

higher speeds, representing the service speed of a road. When retracting data

by means of an inertial profiler, it is essential to carry out investigations at

speeds faster than 15km/h since at lower speeds the operation of the system is

compromised and are not as representative.

Shahin (2005) breaks down the constituents of the inertial profiler in four major

components. These are, a height sensor, necessary to record the height of the

vehicle from the pavement surface, an accelerometer recording the vertical

accelerations of the vehicle, a distance measuring system and a computer.

The acceleration data recorded are processed by double integration, and the

vertical displacement given by this process is then processed by the computer to

give the displacement at each chainage point along the road. This value is

reduced by subtracting the height of the vehicle from the pavement at each

point recorded and the final profile is then presented, usually in terms of IRI or

RV (Shanhin, 2005).

Fig 6. Inertial Profiler (Sayers, Karamichas, 1998)




Marios Papayiannis

10

3 ANALYSIS, DIFFERENCES AND COMPARISSON

3.1 Differences Between Analysis Techniques

Similar to any experimental and analytical technique, the techniques of

measuring profiles are highly influenced by errors. Apart from the risk of human

errors, miscalculations and equipment defects, these techniques have

characteristic idiosyncrasies which can affect their significance and utilisation.

Roughness Variance clearly shows a large difference in values obtained from

specific investigations in New Zealand, compared to those in the UK. For a

specific test the corresponding values for three wavelengths were:

3m wavelength 0mm2-120mm2



These show a dramatic difference from the UK threshold criteria for

maintenance. This difference is not constant but rather ranges from being 10

times bigger for 3m wavelengths to 1000 times bigger for 30m wavelengths.

This is explained as a result of the influence of geometry on the calculation of profile variance and suggests that this should be taken into account, upon

further analysis. Although there is a large difference between these values and

UK threshold values, no other indicator suggests the bad condition of the road

since truck response data do not suggest poor truck ride, according to the UK

discomfort factor criteria (Jamienson, 2008).

There is a strong suggestion by the above data that the UK standards cannot be

applied internationally and that due to other factors there may be necessary tomodify this technique accordingly in an attempt to make it global (Jamienson,

2008).

3.2 Similarities Between Techniques

Benbow, Nesnas and Wright (2006) carried out an investigation where for the

same data, IRIs and profile variance was calculated. In the procedure of

recording the data, passengers were asked to record points were they felt

discomfort and were they sensed the unevenness of the road. This test is acomparison between profile variance, IRI’s and human perception.




Marios Papayiannis

11

Contrary to what was described earlier, there was a good correlation between

the three variables, implying the ability of profile variance to be used globally,

since it results in a common profile to the IRI which has been developed for use

internationally.

Fig 7. Comparison of IRI’s and Profile Variance for the same road

(Benbow et. al, 2006)

As illustrated in figure 7, there is good agreement between results and an

encouraging visual relationship of dial presses, considering that there is

expected delay due to human response time (Benbow et. al, 2006).




Marios Papayiannis

12

4. FURTHER ANALYSIS FOR DERIVATION OF DETERIORATION MODEL

4.1 Construction of Deterioration Model

Following the assessment of experimental information and mathematical

computations, constructing a road profile, is then possible to make direct

decisions concerning the maintenance requirements of a road at that point.

The question which then arises is whether it is possible to manipulate these data

to produce estimation for the behaviour of the road, and to what extent this can

be utilised in order to assess an asset management system and use this data in

order to save assets according to these predictions.

4.2 Effect of Errors on Prediction Models

Producing a prediction model is highly influenced by the quality of the available

data and their abundance, since the ability of producing such a model is

dependent on the quantity and quantity of data.

The ability of an assessor to prepare and analyse historical data by several

methods, will also have a great impact on the error of the estimation and the

extent of reliability of the model as illustrated in figure 8 (Byrne, Parry, 2009).

Fig 8. Error in final estimation (Byrne, Parry, 2009)

Furthermore, any effects of miscalculations and misinterpretations of present

data will have an effect on their future applications. In such case there will be

extra error in the prediction, which will be imposed to the model. Figure 9 shows

the impact of such error on the prediction model. So far prediction models have

been assumed to be linear and the effect of error factors α and β very similar.

This is not the case (Byrne, Parry, 2009).




Marios Papayiannis

13

Fig 9. Effect of present and future errors on prediction model

(Byrne, Parry, 2009)

Deterioration is non-linear, and in such case there will be an additional effect

which will alter the value of α by an extent of x β (where x is a factor) due to the

effect of current data errors in the estimation error. Therefore, it can be argued

that it is much more important, according to what was mentioned earlier, to give

more attention to the errors and miscalculations of current data, rather than

focusing on the competence of a deterioration model, since current data will

have a major effect on the derivation of the result (Byrne, Parry, 2009).

4.3 Other Factors Affecting Deterioration Model

Frequency of measurements has a significant effect on the quality of a

deterioration model. Absence of data, obligates the assessor to extrapolate data

from older records increasing the risk of error occurrence which will be

dependent on the competence of the extrapolation technique and the quality of

the historic data (Byrne, Parry, 2009).

4.4 Minimizing Error

In the process of the computation of a reliable technique for predicting the

deterioration rates of different roads, researchers have produced data mining

criteria, in order to exclude outliers from the data pool. By using such

techniques, the competence of the information retrieved from such model can be

quantified, since there has been an exclusion of outliers and miscalculations.

Therefore it will be possible to produce much more accurate models, reducing

errors and inaccuracies and promote a much more efficient process of assetmanagement (Byrne, Parry, 2009).




Marios Papayiannis

14

Complicated mining algorithms, such as the Minimum Message Length Two

Dimensional Segmenter and advanced computer software have been developed

and are under development. These techniques are responsible for minimizing the

effect of errors on the forecasting procedure. Utilising them will allow for the

most reliable information to be retracted for a given road, according to data

derived from present and past records (Byrne, Parry, 2009).




Marios Papayiannis

15

5 CONCLUSION

5.1 Discussion

Current economic conditions and the rapid change of public expectation from the

road networks in developed countries, have led to the demand of producing a

model which can predict when any road will require maintenance, in advance.

Shortage of budgets and other resources have led governments and other

institutions to promote initiatives which will allow for accurate forecasting road

maintenance requirements and timing.

From the above analysis it is evident that Highway Asset Management can be

highly dependent on specific and complicated techniques which give rise to the

most suitable management of maintenance timing and cost. There has been

extensive research on the ways of analysis and an abundance of techniques is

available to derive decisions based on them. Therefore it is possible to calculate

accurate predictions for the deterioration of a road and plan according to these,

in order to manage a country’s finances and be prepared for maintenance

without jeopardising the economic balance or the safety of the network.

5.2 Proposal

Utilising more than one techniques of data collection, (ie SCANNER and Inertial

Profiler) it will be possible to create a forecasting model which will implement

both the ride quality as perceived by the passenger and the pavement quality,

derived from laser or ultrasonic inspection. By combining the two it is considered

that it will be possible, using analysis techniques and data processing mentioned

earlier, to produce a grading model for the deterioration of a road. This model

will be used as a reference to the deterioration rate of a road and also give an

indication of the effect of rehabilitation or maintenance on a specific road and itsdifferent segments, therefore acting as a planning tool not only for asset

management in terms of maintenance but asset management in terms if

inspection and investigation since it will allow for monitoring using these

complicated techniques when they will be necessary.

Word Count: 2,886




Marios Papayiannis

16

Data Analysis for Pavament Maintenance Asset Managmet

Documents