Keywords

1 Introduction

Developing countries like India have shown an enormous increase in its road network since last century. According to the data of Annual Report 2015, around 100,000 km of roads has been built in last 15 years in India. As soon as the new roads have been developing, there is a major challenge of maintaining the old roads. Moreover, the freshly laid roads are also getting patches of holes in it because of uncertain climate change, use of less durable building materials and cheap technologies for construction.

Potholes have been provably responsible for causing serious hazards and also lead to vehicular damage. Such damaged roads also lead to lowering of highest possible traffic speed causing more congestion, higher fuel consumption, and finally causing pollution. Some tests have concluded that the alteration in road roughness parameter and the fluctuating speed of the vehicle can significantly decrease heartbeat of the driver which impact driver’s response significantly [1]. There is a need of characterizing roughness of the pavement profile so that a proper check of ride comfort as well as safety can be achieved.

IRI is a roughness measurement unit, represented as cumulative vertical distance travel in unit horizontal length. It stands for international roughness index, expressed in mm/km, or in./mi. It describes road roughness used for evaluation and management of road system. It is reproducible, stable, and applicable for every terrain. IRI measuring devices are categorized into following classes on the basis of its accuracy:

  • Type 1 profiler—these give exact profile with high precision;

  • Type 2 profiler—these are the non-biased type of profilers, e.g., Merlin;

  • Type 3 profiler—these are the response-type devices with 80% accurate results, e.g., smartphone apps;

  • Type 4 profiler—these are the least accurate profiler based on subjective ratings.

IRI is correlated with the following variables:

  • Pavement characteristic (a statistical quantity)

  • Vertical vibration (pavement quality)

  • Tire load (safety and control ability) [2].

Historically, there has been various equipment and techniques for determining pavement quality. One of such equipment is a towed fifth wheel bump integrator that is very popular in developing countries like India. Its composition is same as quarter car model dynamically, where the spring system supports the upper part and the wheel helps to protect against unexpected excitations, but this is really a tedious and expensive technique and the instrument is too heavy to handle. Similarly, dipstick can be used for determining IRI and also for measuring road roughness, but cracks cannot be sensed through it. While Merlin Cycle works on the speed of a cycle towed by hand, so the data collection is again a time-taking phenomenon. Conclusion is that every profiler has its certain limitations.

One of the emerging technologies to find IRI is based on Android-based smartphone applications. Smartphones are the mobile applications equipped with the technologies like global positioning system (GPS), sensor, gyroscope, data collection and storing potentiality, image-capturing potentialities, and fast processability. These features enable it to be used as a pavement surface classification instrument. In this paper, the variation of the roughness value, that is, IRI value obtained using smartphone at a varying speed and vehicle type is being discussed and correlated. The standard value of IRI is obtained using standard profiler, that is Merlin equipment, and comparison is made.

2 Literature Review

Many research works have been done in pavement profiling since nineteenth century. Most of those techniques proved to be reliable in estimating unevenness. Some of the research works related to pavement profiling are summarized below.

2.1 Literature Review on Profiling Parameters

Michael W. Sayers, Thomas D. Gillespie, and Cesar A. V. Queiroz (1986) characterized the pavement unevenness in a universal, consistent, and relevant manner and evaluated standard indices based on the geometric characteristics, road simulation, vehicular characteristics, and spectral analysis of the roughness recorder output. Response-type pavement unevenness measurement system based on vehicle simulation was used to calibrate profile [3].

Michael W. Sayers (1998) wrote a little book of profiling for understanding and determining road profile. In this book three basic questions were answered, that is, (1) How profilers are used? (2) How it helps? (3) How we can reduce errors? He described three basic parameters of a profiler, that is, a certain elevation, a height with respect to that elevation, and a horizontal length, where rod and level were defined as static profiler and dipstick as a dynamic profiler. He proposed the use of power spectral density (PSD) function as a high-speed profiling system, which is based on the classification of voltage, and applied the same mathematical function for profile measurement. He also elaborated that longitudinal acceleration and vertical vibration experienced by the rider while moving in a high seat vehicle like truck is higher than that experienced by the rider of a passenger car due to road roughness leading to a situation of discomfort.

Peter Mucka and Johan Granlund (2012) dealt with estimating the effect of the contents of the wavelength on IRI. It was found that the IRI corresponding to a velocity of 20 kmph is twice sensitive to velocity of 80 kmph for a local obstacle [4].

Peter Mucka (2016) proposed velocity-related IRI limit curves. He observed the large range of RMS values on the basis of vibrational response corresponding to an IRI value and quantified comfort level and safety of the ride along with the dynamic pressure [5].

2.2 Literature Review on Different Profiling Techniques

M. A. Cundill (1991) devised a simple low-cost pavement roughness data measuring machine, that is, MERLIN, which stands for a machine for evaluating roughness using low-cost instrumentation. This device has been well correlated with the pavement roughness data measuring machine known as bump integrator. Also the roughness output obtained is correlated to get the roughness value in terms of IRI (m/km).

Oldrich Kropac and Peter Mucka (2009) proposed an indicator defining about the pavement unevenness on the basis of the response of the vehicle on vertical vibration caused by the waviness of the pavement profile. Also they proposed modified IRI values which are affected by speed, and assessment of subjective rating methods [6].

Manish Paul and Rumi Sutradhar (2014) derived a generalized equation to get the values of IRI using bump integrator at any speed corresponding to 32 kmph speed by using SPSS software [7].

$$\left( {\text{BI}} \right)_{32} = 0.956\left( {\text{BI}} \right)V + \, 0.842V{-}25.544 \, \left( {R^{2} = 0.958} \right)$$
(1)

Marwan Hafez, Khaled Ksaibati, and Richard Anderson-Sprecher (2016) used statistical technique and old PMS data and then developed uni-variable regression multiple accusation. It was concluded that by the use of historical data, a good estimation of pavement data can be obtained [8].

2.3 Literature Review on Smartphone-Based Profiling Technology

Kasun De Zoysa, Kasun Chamath, Keppitiyagama chamath (2007) designed a road surface observing system which worked on a sensor-based network Bus Net. Bus Net is an ideal approach to monitor data network using a public transport because public transport uses the road that we want to monitor and the most economical attempt for monitoring road condition [9].

Shahidul Islam, William G. Buttlar, Roberto G. Aldunate, and William R. Vavrik (2014) collected IRI values at two test sites using smartphone technology and validated those values with the help of values of standard inertial profiler. They found that in 37 out of 40 tests, values were within 15% of the standard results. A linear correlation can bring the close result which can be implemented, if required [10].

Trevor Hanson, Coady Cameron, and Eric Hildebrand (2014) calculated IRI values from different smartphones to compare the value of IRI by varying type of device, vehicular speed, type of vehicle, and also mounting arrangement with respect to a class-1 profiler.

They concluded that significant factor causing variation in IRI values are type of smartphone used, mounting arrangement, and type of vehicle [11].

Rajiv Kumar, Abhijit Mukherjee, and V.P. Singh (2016) performed crowd sourcing, that is, distributed smartphones to the people for gathering road roughness parameter; used a fuzzy system and characterized the roughness parameter; presented road surface condition on web mapping service platform by different patterns of surface classes; and the roads were monitored visually for justification of the smartphone technology [12].

3 Objective and Scope

It is already concluded above that the emerging smartphone technologies determine pavement profile in a very short time. So, the objective of this paper was to detect pavement unevenness of a certain test section using smartphone application and correlating the obtained IRI values with the most reliable value of IRI obtained using a standard equipment at the same section. The smartphone values were determined at a varying condition of speed and type of bike used to understand the variation in IRI values with the respective variations. The standard equipment used was a type-2 profiler, that is, Merlin Cycle.

4 Methodology

In this paper, before performing any experiment, some basic components of the experiment need to be understood well. Following are the basic requirements:

4.1 System Architecture

System of pavement roughness data analysis consists of the following parameters:

Smartphone: Smartphone embedded with Android OS > 5.0 is to be taken, so in this paper MOTO G5 Android version 8.1.0 had been taken as a smartphone platform.

Smartphone Application: ROADROID mobile application as shown in Fig. 1 installed in smartphone with the following features in it:

Fig. 1
figure 1

ROADROID application installed in smartphone

Full size image

  • Analyze vehicle vibration in 100–200 Hz

  • Calculate two IRI values, that is, calculated IRI (CIRI) and estimated IRI (EIRI) and GPS

  • Sensitivity and segment length adjustment

  • Degree of accuracy 80% of type-1 IRI measuring equipment

  • ROADROID is a type-3 IRI measurement equipment, that is, a response-type survey system

  • EIRI value is calculated from quarter car formula

  • CIRI value uses a smothering filter, hence reported as the required IRI [13].

Vehicle: In this study, two motor bikes have been selected for performing experimental study. One of them is shown in Fig. 2 (Table 1).

Fig. 2
figure 2

Image of the smartphone mounted on bike

Full size image
Table 1 Profile of test vehicle
Full size table

Survey Speed: A constant speed needs to be maintained as the speed considerably affects the vertical vibration. In this paper, two speeds were maintained and values had been recorded in 25 and 30 kmph speed.

Test Section: Since Merlin equipment was available in NIT Raipur campus; therefore, a 400 m test section was taken from NIT Raipur campus, as shown in Fig. 3.

Fig. 3
figure 3

Test section snapshot taken by using Google Map

Full size image

Software: Using SPSS software, data obtained from experimental results are represented graphically and a correlation between different observations is made. A snapshot is shown in Fig. 4 [14].

Fig. 4
figure 4

Data view in SPSS software (snapshot)

Full size image

Mounting Arrangement: A stable mount is necessary to get an accurate profile and vertical vibrations.

Merlin Equipment: It is a class-2 profiler which gives accurate IRI value. MERLIN stands for machine for evaluating roughness using low-cost instrumentation. It consists of a graph paper in which displacements are plotted as a histogram. A working equipment has been shown in Fig. 5. In this equipment, the probe is connected to an arm such that it is moving close to the probe. To the next side of the arm, a pointer is attached such that 1 mm movement of the probe moves the pointer by 1 cm over the prepared data chart. The chart consists of column divided into 5 mm boxes [15].

Fig. 5
figure 5

Merlin equipment while working (author Mandeep Kaur Arora along with co-author Mahesh Ram Patel recording data)

Full size image
Fig. 6
figure 6

A snapshot of downloaded aggregated file (obtained through smartphone application)

Full size image

4.2 Data Collection

Using Smartphone App: This was the first step of pavement roughness measurement in this paper. As shown in Fig. 2, the smartphone, that is MOTO G5, was stably mount in the handle of the bike. Such a fixing was done so that there was no self-movement of the smartphone. Bike was ridden at a constant speed and the recording of the values was carried out in smartphone app simultaneously. In this experiment two values of speed, that is 25 and 30 kmph, were taken for recording values. Recording was done in both of the bikes separately, one at a time. Specifications related to application, smartphone, and motor bike are already mentioned in the above section. Finally, the readings were uploaded from the application which was opened further by logging in ROADROID website for getting the readings in terms of IRI.

Using MERLIN Equipment: To measure roughness of road using MERLIN equipment, 200 observations were made. Each observation was taken by resting the machine on road when the wheel was in stable position and the probe, stabilizer, and back end foot in contact with the pavement. Then the pointer was recorded in the graph and a tally box with a cross-sign in the respective column is shown in Fig. 7, to keep an observation of the records. Then the handle was elevated such that the wheel was only in contact with the pavement surface and then was taken ahead for the next normal position and the same procedure was repeated. After 200 observations were made, chart was removed from the MERLIN.

Fig. 7
figure 7

Chart obtained from Merlin equipment

Full size image
Fig. 8
figure 8

a Correlation in IRI values in Honda LIVO. b Correlation in Honda CB Shine

Full size image

4.3 Data Analysis

Smartphone Data: The downloaded data are available in .txt format, as shown in Fig. 6. The readings corresponding to CIRI were exported to SPSS software and average IRI was calculated and also a graphical representation of the roughness was plotted as shown in Fig. 9. Further, there was also a correlation and regression analysis made between two values of IRI corresponding to 25 and 30 kmph speed, using SPSS software.

Fig. 9
figure 9figure 9

a IRI versus distance @25 kmph speed in Honda LIVO. b IRI versus distance @30 kmph speed in Honda LIVO. c IRI versus distance @25 kmph speed in Honda CB Shine. d IRI versus distance @30 kmph in Honda CB Shine

Full size image

MERLIN Data: The chart obtained after the experiment was observed and the position from the both sides of the chart after 10th cross was marked. The distance between the two cross was measured in mm and calculated as D mm, that is, 45 mm in this experiment. This value was the roughness in terms of MERLIN scale. Then pavement unevenness in terms of IRI was obtained with the help of this relation

$${\text{IRI}} = 0.593 + 0.0471D$$
(2)

5 Flow Chart

figure a

6 Results and Discussion

6.1 Merlin

The standard value of IRI obtained from Merlin equipment was 2.7125 m/km.

6.2 Smartphone and Software

The test results demonstrate that there is a significant variation in average IRI value obtained when the speed is increased. Following are the results corresponding to both the bikes:

Honda LIVO: In this bike, the value of IRI obtained @25 kmph speed was 2.9330 m/km, which was close to the standard value, that is, within 10% of the standard value. However, the values @30 kmph speed was 4.6647 m/km, which was highly deviated from the standard IRI, that is, differ by 71.97% from standard IRI.

Honda CB Shine: In this bike, the value of IRI obtained @25 kmph speed was 3.0032 m/km, which was again close to the standard value, that is, within 10% of the standard value. However, the values @30 kmph speed was 4.0352 m/km, which was once again highly deviated from the standard IRI, that is, differ by 48.763% from standard IRI.

This can be seen that the variation in vehicle type does not significantly impact because the suspension type and chassis type are almost similar in both the cases. Only variation is in engine of the vehicle, which is not causing large variation in values (Table 2).

Table 2 Variation in IRI value by varying speed and vehicle type
Full size table

6.3 Correlation and Regression Analysis

Using SPSS software, there was a correlation and regression analysis established between IRI values obtained @25 kmph values with that of 30 kmph values, such that 25 kmph values were the independent variable x and 30 kmph values were dependent variable y. Positive value of correlation represented in Table 3 of Honda LIVO bike represents an increase in difference of IRI value with increase in speed.

Table 3 Correlation in IRI values @25 and 30 kmph speed obtained from Honda LIVO
Full size table

Standard deviation is 0.986. The required relation is

$$y = 3.32 + 0.46x$$
(3)

However, a negative correlation shown in Table 4 was found in Honda CB Shine values, which represents that on increasing value of speed in this bike, the difference in IRI value decreases. Standard deviation is 0.987. The required relation is

Table 4 Correlation in IRI values @25 and 30 kmph speed obtained from Honda CB Shine
Full size table
$$y = 4.55 - 0.17x$$
(4)

Figure 8 shows the correlation in both the vehicles. Figure 10 shows the histogram. Figure 11 shows the normal P–P plot of regression standardized residual.

Fig. 10
figure 10

a Histogram in Honda LIVO. b Histogram in Honda CB Shine

Full size image
Fig. 11
figure 11

a Normal P–P plot of regression standardized residual in Honda LIVO. b Normal P–P plot of regression standardized residual in Honda LIVO

Full size image

7 Conclusion

This paper demonstrated the capability of a smartphone to measure pavement profile with the help of its in-built feature accelerometer. Use of a smartphone app, that is ROADROID, visual interface system, that is MS Excel, correlation and regression analysis software, that is SPSS, has enabled accelerometer feature of a smartphone to give a comparable pavement surface unevenness measurement. In this paper, two motor bikes, Honda CB Shine and Honda LIVO, were used with a suitable mount fixed in it to give a platform to smartphone, that is, Moto G5 for measuring values of IRI in a 400 m stretch of road. The values were recorded at a varying speed of 25 and 30 kmph in both the motor bikes. For the validation of obtained results, a standard profiler, that is Merlin Equipment, was used to record the pavement unevenness value in terms of IRI. IRI value obtained through Merlin equipment was 2.7125 m/km. IRI value obtained through smartphone was within 10% of this value when measurement was made @25 kmph, whereas measurement made @30 kmph gave a significant difference in IRI value, that is about 48 and 72%, respectively, in both the vehicles. A comparison of these values yielded the following conclusion:

  • It has been assured that the pavement profile is easily obtainable with the help of smartphone equipped accelerometer technique.

  • Higher value of speed gives higher value of IRI, because of higher vertical vibration

  • Varying motor vehicle of same kind does not significantly cause the variation in IRI, if the suspension system is not varying

  • In Honda LIVO bike, with increase in speed, difference in IRI value also increases.

  • However, in Honda CB Shine with increase in speed, difference in IRI value decreases.

  • More accurate profile of pavement is obtained @25 kmph speed

  • IRI value obtained @ 30 kmph speed in Honda LIVO can be approximated to IRI value of 25 kmph by using relation y = 3.32 + 0.46x

  • IRI value obtained @ 30 kmph speed in Honda CB Shine can be approximated to IRI value of 25 kmph by using relation y = 4.55 − 0.17x

  • IRI value obtained from smartphone app is generally higher than the standard equipment because of the engine vibration of vehicle.