...

Development of methods and tools for monitoring and analyzing customer data

by user

on
Category: Documents
2

views

Report

Comments

Transcript

Development of methods and tools for monitoring and analyzing customer data
Development of methods and tools for monitoring and analyzing customer
data
Author
Shashank Velpanur
Master Thesis at
Department of Management and Engineering (IEI)
Department of Quality Technology and Management
Supervisors at Volvo Construction Equipment: Dr. Peter Wallin & Dr. Jonas Larsson
Supervisor at Linköping University: Dag Swartling
Examiner at Linköping University: Lars Witell
Degree Project
Department of Management and Engineering
LIU-IEI-TEK-A--13/01561--SE
Abstract
The market in today’s world is dominated by customers and their requirements. Customer
feedback is essential for manufacturing companies trying to establish a foothold in the service
environment. New product development is often recognized as the trump card held by most
manufacturing companies that allow them to gain a competitive edge over their rivals in the
market. Utilizing customer’s feedback in new product development is a step that many
companies are taking to be able to satisfy customer needs and requirements which enables the
companies to firstly, retain the customers and secondly, bring in new business. Interviewing
customers for feedback is a very common method employed by service companies to be able to
capture and store data which is then analyzed to identify any patterns emerging i.e. any particular
features preferred, any features disliked, etc. This thesis uses interviews as a method to be able to
capture feedback from the customers which can then be utilized in new product development.
The basis of this thesis is formed by developing a method for analyzing and monitoring customer
data. The customer perceives a product or a service based on his/ her experience in using it and
hence forms an opinion on it. This thesis mainly focuses on interacting with internal operators
(Volvo CE employees who participated in a previous measurement) to understand and collect
their feedback of the product, the L220F wheel loader, as an example to develop the method, due
to the fact that “real” customers are not as easy to reach and interview. The data collected will
then be compared to logged data of the product usage by these operators. Finding a correlation
between the answers to the interviews and the measured data will help identify the gaps in how
the wheel loader is used by operators of different skill levels and how it can be used. If
successful, this method can then be utilized on external customers of Volvo CE and also on other
products manufactured by Volvo CE.
The conclusions drawn from this thesis are that while all customers must be equally considered
while taking feedback, whether they are professionals or rookies, Volvo CE should rate their
answers differently while designing a new product to meet customer requirements. Also, what
could be seen from the clustering is that more heterogeneous groupings of operators are formed
wherein no one cluster is made up of purely one class of operators.
I
Acknowledgement
The thesis work is a part of my Master’s degree in Industrial Engineering & Management at
Linköping University and is accomplished at Volvo Construction Equipment, Eskilstuna,
Sweden. On this occasion I would like to extend my heartfelt gratitude to Volvo CE and the
Advanced Engineering Department who supported me throughout the project.
Foremost, I would like to thank Dr. Rikard Mäki, Director, Technology Planning & Public
Funding, for giving me this opportunity to be part of his department and for taking an active
interest in the progress of this thesis.
I am grateful to my mentor Dr. Peter Wallin, Research Co-ordinator and my supervisor Dr. Jonas
Larsson, Chief Project Manager, who have supported me and spent their invaluable time in every
aspect of this thesis. Their constant feedback and knowledge have helped me to solve many
problems at critical junctures.
I would like to extend my heartfelt thanks to Mr. Bobbie Frank, Alternative Drivetrain Research
Engineer, who has spent a lot of his time guiding me through different aspects of this thesis and
without whose knowledge, it would not have been possible to complete this report.
Importantly, I would like to take this opportunity to thank Mr. Jostein Langstrand, Director of
Studies, Quality Technology and Management, Department of Management and Engineering,
Linköping University for accepting this thesis as part of his department and for his support in the
completion of this thesis.
I would also like to extend my gratitude to my examiner Mr. Lars Witell, Associate Professor
and my supervisor Mr. Dag Swartling, Quality Technology and Management, Department of
Management and Engineering, for their ideas, knowledge and support right from the beginning
of this thesis.
Finally, I would like to extend a heartfelt thanks to my family and my girlfriend, Ms. Sharika
Pisharasyar, without whose unending support and encouragement, it would not have been
possible to accomplish this project and thesis report.
II
Contents
1 Introduction ................................................................................................................................................ 1
1.1 Corporate Overview - Volvo Construction Equipment ...................................................................... 1
1.2 Introduction to thesis subject .............................................................................................................. 3
1.3 Background ......................................................................................................................................... 4
1.4 Purpose and Objectives ....................................................................................................................... 4
1.5 Delimitations ....................................................................................................................................... 5
2 Theoretical frame of reference ................................................................................................................... 6
2.1 Importance of customer feedback in new product development......................................................... 6
2.2 Methods to collect customer feedback ................................................................................................ 8
2.3 Satisfying internal customers before handling external customers ..................................................... 9
3 Methodology ............................................................................................................................................ 10
3.1 Research method ............................................................................................................................... 10
3.1.1 Choosing interviewees from the organization............................................................................ 10
3.1.2 Interview structure (including coding of answers received) ...................................................... 10
3.1.3 Analyses of answers received from interviews .......................................................................... 11
3.1.4 Clustering or grouping customers based on their answers ......................................................... 13
3.2 Background of tests conducted in September 2011 .......................................................................... 15
3.3 Interview Process .............................................................................................................................. 16
4 Analyses of Results .................................................................................................................................. 17
4.1 Engine Speed during bucket fill phase with fuel efficiency ............................................................. 22
4.1.1 Short Loading Cycle (Gravel) and Load & Carry (Gravel) ....................................................... 23
4.1.2 Short Loading Cycle (Rock) ...................................................................................................... 26
4.2 Vehicle Speed during transport ......................................................................................................... 33
4.2.1 Short Loading Cycles (Gravel & Rock) ..................................................................................... 34
4.2.2 Load & Carry (Gravel) ............................................................................................................... 38
4.3 Cycle Distance .................................................................................................................................. 42
4.4 Lifting force during the bucket fill phase.......................................................................................... 51
4.5 Engine Power .................................................................................................................................... 56
4.6 Cycle Times ...................................................................................................................................... 68
4.6.1 Cycle Times for Short Loading Cycle (Gravel) ......................................................................... 69
4.6.2 Cycle Times for Load & Carry (Gravel) .................................................................................... 73
III
4.6.3 Cycle Times for Short Loading Cycle (Rock) ........................................................................... 74
4.7 Fuel Efficiency .................................................................................................................................. 79
4.8 Rimpull ............................................................................................................................................. 84
5 Clustering ................................................................................................................................................. 90
5.1 Clustering of accuracy (difference between answered and actual values from Chapter I) ............... 91
5.2 Clustering of preferences from chapter II ......................................................................................... 93
5.3 Clustering of parameters from chapter II using weighted scores ...................................................... 95
5.4 Clustering of actual values of parameters from chapter II ................................................................ 97
5.5 Clustering of combination of preferences and actual values from chapter II ................................... 99
6 Conclusions ............................................................................................................................................ 101
7 Future Works ......................................................................................................................................... 104
8 References .............................................................................................................................................. 105
Appendices................................................................................................................................................ 109
IV
Acronyms
VCE
Volvo Construction Equipment
WLO
Wheel Loader
NPD
New Product Development
RPM
Revolutions per Minute
SLC
Short Loading Cycle
L&C
Load and Carry
ICE
Internal Combustion Engine
IP
Internal Professionals
IA
Internal Average
IR
Internal Rookie
QFD
Quality Function Deployment
CSM
Customer Satisfaction Modeling
V
1 Introduction
1.1 Corporate Overview - Volvo Construction Equipment
Volvo Construction Equipment (VCE) is one of the world's largest manufacturers of construction
machines, with a full product range, manufactured, serviced and supported all over the globe.
Volvo CE offers products and services in more than 125 countries through proprietary or
independent dealerships. Volvo CE’s products are used in energy related industries (oil & gas),
road construction, building, demolition, industrial material handling, recycling industries, to
name a few.
Volvo CE’s product range includes wheeled and crawler excavators, articulated haulers, scraper
haulers, wheel loaders, pipe layers, demolition equipment, waste handlers, motor graders, pavers,
compactors, milling equipment, tack distributors, road wideners, material transfer vehicles and a
range of compact equipment such as backhoe loaders and skidsteer loaders.
Volvo CE has a range of wheel loaders from small size L15 variant to a large size L350 variant.
The wheel loader L220 belongs to the large machine category. The fields of applications are civil
& building construction, waste handling, recycling, log handling, pallet fork operation etc.
(www.volvoce.com)
Figure 1: Volvo CE wheel loader in operation loading blasted rocks
1
Figure 2: Volvo CE wheel loader in operation loading sand
The 3 main applications investigated during this thesis are:1. Rehandling (Gravel) – Short Loading Cycle (Gravel) called Taltet
2. Rehandling (Gravel) – Load & Carry Cycle (Gravel) called Bandet
3. Primary Extraction from Face (Blasted Rock) - Short Loading Cycle (Rock) called Berg
2
1.2 Introduction to thesis subject
Customer feedback in new product development is a very important issue in building
relationships for a manufacturing company’s success (Gruner and Homberg, 2000). The ultimate
measure of new product performance is customer satisfaction which influences the product’s
success or failure in the market. Ignoring customer feedback leads to dissatisfied customers and
ultimately poor quality products which never fulfill the requirements of the end user, resulting in
large losses to manufacturing companies. Rather than focus on what is wrong in an existing
product, companies should shift their focus to what the customer really wants, incorporating their
needs into new product development (Bouchereau and Rowlands, 2000).
The requirement of this thesis is to develop a methodology that combines the customer
preferences with their actual usage of the product. The feedback received from the customers
(internal employees) can then be utilized in designing better products and thereby satisfy the
requirements of the customer. This allows Volvo CE to be in a better position to serve their
external customers as well.
Performance measurement is the true reflection of the product’s ability to meet customer
demands. Product performance measurements help the manufacturing company to monitor,
control and improve the product’s characteristics and parameters continuously, thereby
producing a product that satisfies the customer’s requirements (Kollberg et al., 2005; Elg, 2007;
Bergman and Klefsjo, 2003). By finding correlations between different measurable and
parameters, gaps in the product usage can be identified and eliminated. Neely et al. (1995)
identified quality, costs, time and flexibility as the four main factors along which manufacturing
companies must align themselves to be able to efficient and productive. Designing,
implementing and using performance measurement techniques largely depends on the product
parameters and characteristics being measured.
Burnard (1991) describes a method of analyzing data from interviews as a 14 stage process,
wherein the interviews carried out are semi-structured and open-ended, thereby allowing the
questionnaire are codified by taking actual values for some questions and ratings for the others.
The analysis done for each question includes the cleaning of data which is a process during
which the data is inspected and any wrong data is corrected, if possible (Adèr and Mellenbergh,
2008). In their book on research methods, Adèr and Mellenbergh (2008) propose a method for
3
data analyses which includes the checking of the quality of data, checking the quality of
measurements, performing any initial transformations and finally, implementing the
transformations in the analyses. Once the data has been sorted and found to be usable, the main
data analyses can be done and the stability of the results should be checked by methods such as
cross-validation or sensitivity analysis. Using popular statistical methods such as ANOVA can
help validate the results of the analysis, in their opinion.
The final part of this report deals with data clustering. Cluster analysis is widely used in many
fields such as statistics, social sciences, biology, manufacturing and software engineering to be
able to identify groups of customers based on large amounts of data. Most manufacturing
companies perform clustering on their customers to be able to design specific products for
specific market segments based on their requirements. In short, cluster analysis groups data
objects into groups such that objects that belong to the same cluster are similar, while those that
belong to different clusters are not. There are many definitions of clustering. Jain and Dubes
(1988) define clustering as follows:
“Cluster analysis organizes data by abstracting underlying structure either as a grouping of
individuals or as a hierarchy of groups. The representation can then be investigated to see if the
data group according to preconceived ideas or to suggest new experiments”.
1.3 Background
The basis of this thesis is formed by analyzing and developing a method that enables improving
the product performance characteristics through customer feedback. The conversion of
monitored data into useful knowledge in the process of product development is necessary to be
able to design the product as per the customer’s requirements. Customer preferences,
requirements and how the customer uses the product are all knowledge that helps in designing
the right product.
1.4 Purpose and Objectives
The purpose and objectives of the thesis define the goals of the thesis. The goal is the end result
that is expected to come out of the thesis project. A clear problem statement allows for a
4
structured way to identify what needs to be done and the steps taken to accomplish the same. The
initial problem statement and the goal of the thesis are set by discussions between the thesis
author and his mentor and supervisor at Volvo CE.
One of the main goals of this thesis is to develop a method to combine operator preferences with
actual usage measurements to impact the design requirements. This should be accomplished with
support from literature and interviews.
Once the feedback from the interviews has been taken, the next step is to develop a system to
identify what measurements are useful to evaluate from the data collected from interviews and
product measurements captured from the CAN-bus signals.
Finally, to analyze the collected data to identify patterns and cluster the operators into categories
according to how they value the various product characteristics. This clustering of the operators
will help Volvo CE to develop products to suit particular customer segments according to their
requirements.
1.5 Delimitations
The delimitations of this thesis are the areas that the thesis is restricted to. While it is possible to
have a very wide field of study and analysis, the time allocation for the completion of the project
limits the thesis to certain areas as highlighted below:
-
This thesis is limited to the study of product characteristics of the L220F Volvo CE wheel
loader. If the interview methods are found to be successful and yield measurable data,
this method can then be extended to the other variants of the wheel loaders as well as to
other Volvo CE products.
-
The interviews are carried out only within the internal group of operators (test engineers,
design engineers, operators, etc.) and not with the external customers (though the scope
of the thesis can be extended to the external customers if found to be necessary).
5
2 Theoretical frame of reference
Based on extensive literature study done, guided by the gained educational knowledge, feedback
from the university supervisor and contacts at Volvo CE, the important areas in this thesis project
have been identified and singled out. The purpose of this frame of reference is to gain a
theoretical background into the main areas of customer feedback process and its benefits for a
manufacturing company.
2.1 Importance of customer feedback in new product development
Many companies rely on their customers to give feedback and make use of their competencies in
new product development. Kristensson et al., (2011) highlight companies such as Dell, P&G and
Google as such companies that value customer co-creation in their innovation process. They
argue that customer co-creation is becoming increasingly popular among companies which
ultimately determine the success of a new product. In their paper on collaborating with
customers, they analyze the interaction with the customer based on four dimensions – frequency,
direction, modality and content – to better understand the value of customer collaboration. Witell
et al., (2011) successfully argue that new offerings in the market that are developed through
market research methods based on collaborating with customers are far more successful and
profitable than those that are developed through more traditional customer feedback and market
research methods.
The literature study done in this field categorizes customers as passive or active; depending on
whether companies utilize a proactive or reactive market response method (Narver et al., 2004).
Narver et al., (2004) also suggests that companies that utilize proactive market approaches tend
to work more closely with their customers. Proactive market response is a process driven by the
customer where the companies must discover, understand and fulfill the needs of the customer
by closely working with them and involving them with market research techniques. This will
help the companies to satisfy the customer’s current needs as well as predict the future needs. It
also helps the companies to discover new market opportunities (Jaworski et al., 2000). Reactive
market response on the other hand enables the company to fulfill those needs that are expressed
by the customer.
The idea of co-creation is proposed by many authors in their papers (Grönroos and Ravald, 2011;
Wittel et al., 2011; Payne et al., 2008; Vargo, 2008). The implications of co-creation are that
6
customers are involved in creating value in a product only during the consumption phase. But as
many authors argue, the customers should also be involved in innovation phase so that their
feedback can be utilized in designing a better product or service (Kristensson et al., 2004; Alam,
2002).
Figure 3: Value Creation Spheres (Grönroos, 2010)
In his paper on value co-creation, Grönroos (2011) illustrates the roles of the company and the
customer, as shown in figure 3, in the co-creation process. He highlights the fact that in the
production sphere, the company is responsible for the production process wherein, the company
produces the resources to be used in the customer’s creation of value. In the joint sphere
(interaction sphere), the customer co-produces the resources and processes with the providing
company and is ultimately the value creator. In the rest of the customer sphere, the customer
creates value from the product or the process independently from the provider, as this sphere is a
closed sphere from the provider. Since, this sphere is a closed system, the providing company
can not take part in the customer’s experience of the product or service in use. Companies
striving to be able to understand and incorporate the customer’s feedback and ideas in the cocreation process, should invite the customer to participate more in the value creation process,
7
thereby allowing the customer to influence the design and development of the product or service
(Grönroos, 2011; Voima et al., 2011).
2.2 Methods to collect customer feedback
Some of the more commonly used market research methods used by companies are surveys,
focus groups, comment cards, and in-depth interviews (Verma et al., 2008). These methods aim
to capture the customer’s previous experiences with a product or service and are designed such
that the customer responds to them when asked by the company. These methods are therefore
classified as reactive methods (Johnson, 1998). Wittel et al., (2011) argue that in these reactive
methods, the companies pre-determine what questions to ask, for example in in-depth interviews
or surveys, or they constrict the customer responses to their previous experiences of utilizing the
product or service. In such scenarios, the customer is more likely to respond only based on
his/her usage experience and not really be able to give new ideas or allow the company to build a
profile for future usage of the product (Trott, 2001).
The main difference between a proactive and reactive customer feedback method (Wittel et al.,
2011 via Johnson, 1998) is that proactive methods capture the latent needs of the customer, both
from previous usage experience and those for future usage whereas reactive methods tend to
capture only the feedback from customers, based on previous experiences with the product. The
method utilized in this thesis, in-depth interviews, are cited as being able to capture only the
spoken needs of the customer (Gustafsson et al., 1999) and that there are better methods to
capture the latent needs of the customer.
Using interviews as a method to collect customer feedback is an effective way to be able to get
firsthand accounts as to how the product works and reacts under different conditions. Customers
are often able to propose their ideas as to how they want the product to work. By capturing and
storing the feedback, it can then be analyzed to identify patterns or clusters related to preferable
product characteristics and different customer segments. Therefore, to be able to capture both the
spoken and unspoken needs of the customer, the questionnaire for the interviews in this thesis
was designed by taking the customer’s perspective and feedback.
8
2.3 Satisfying internal customers before handling external customers
For any company to successfully satisfy its customers, it should also consider the internal
customers while designing a product or service. The below quote by Benjamin Schneider,
adopted by Smith (2012) in his book on the link between individuals and organizations,
highlights the link between internal and external customers as not being as tenuous as one might
think:
“There’s a remarkably close and consistent link between how internal customers are treated and
how external customers perceive the quality of your organization’s services. A commitment to
serve internal customers invariably shows itself to external customers. It’s almost impossible to
provide good external service if your organization is not providing good internal service.”
Gounaris (2008) emphasizes that it is important to satisfy the needs of the employees before the
company can start thinking about the external customers. Roberts-Lombard (2010) is his paper
on treating employees as customers highlights the fact that the satisfaction of employee needs, as
internal customers, allows the company to be in a better position to deliver the desired quality of
the product or service to satisfy the external customers. Many authors like Gounaris (2008),
Lucas and Kline (2008), Roberts-Lombard (2010), argue that the success of a business is
impacted by the way in which a company deals with its employees as they adopt the role of the
customers and will have the same requirements as an external customer.
One of the problems that can arise from using internal customers are that some of the employees
may not give honest answers to the questionnaire or interviews, fearing that a negative response
can be linked to them, which will jeopardize their position in the company. The employees will
therefore gloss over the real problems with the product and instead choose to give a safe answer
which may not be helpful while designing the product or service. This is especially more
noticeable among the employees from the blue collar group, as they fear that giving a critique of
the product would endanger their job. This can be circumvented by assuring the interviewees of
confidentiality, which will allow them to express themselves more freely with honest answers
(De Vaus, 2002).
9
3 Methodology
3.1 Research method
The methodology followed during this thesis has been decided after the literature review done by
the thesis author, along with ideas and feedback from the mentor and supervisor at Volvo CE and
Linköping University. Since the interview method used here is a reactive method of collecting
feedback from the internal operators, it is a one-time process which if found successful can then
be extended to the external customers as well. Many authors like Gill et al. (2008), propose
interviews as a method useful mostly in qualitative research, but since it was a method that
would allow for one-on-one interaction with the participants in the previous study, it was chosen
for this thesis as well which deals mostly with quantitative data analysis.
By identifying the similarities and differences between the customer requirements of the product
and the characteristics of the product offered by VCE by relating customer feedback to product
design characteristic, it is possible to identify if there are connections between customer
preferences and how they use the product.
3.1.1 Choosing interviewees from the organization
The first task is to identify a pool of operators to interview to obtain data about their preferences
with respect to product performance and characteristics. The interviews to be conducted should
include all the operators who participated in the previous measurement. Denscombe (2007) and
Hirschman (1986) are two of the many authors who suggest choosing people from different
levels within the organization, to increase the credibility of the study being conducted. Once this
is done, a related literature survey is conducted to prepare lists of interview questions related to
product characteristics and performance measures.
3.1.2 Interview structure (including coding of answers received)
Yin (2008) claims that interviews should be like guided conversations rather than like a
structured query. This is true when the data being analyzed is qualitative, but for this thesis, since
the data is quantitative, the interviews are structured and hence in the first part of the interviews,
the respondents have to answer with a numbered value or with a rating. The second part of the
interview is more open ended and allows the respondent to give his opinion on the Volvo CE
WLO more freely. The key to interviews in his opinion is to follow a line of inquiry and ask
questions in an unbiased manner. The questions asked should be done in friendly and non10
threatening, open ended manner to ensure good responses from the interviewees. In the
interviews conducted, the interviewees were assured that their responses and data collected from
the interviews would be kept confidential, details such as interviewee name and age, for
example, which in De Vaus’s (2002) opinion encourages participation on behalf of the responder
and also increases the quality and honesty of their answers. To ensure that the interviewees were
comfortable with the questions and the environment, the interviews were conducted in the
natural working environment of the interviewee, with the presence of a translator in a couple of
cases.
The rating scale used in the tangible part of the interview was a scale of 1 to 4 wherein 1 & 2
were negative responses and 3 & 4 were positive responses, as shown later. The rating of 0 (for
unsure or no answer) was also included, but none of the questions asked evoked this response
and hence has not been focused upon. The reason the rating scale of 1 to 5 was not done was
because of the thesis author’s own personal experiences with such rating scales wherein
respondents usually choose the middle answer of 3, generally associated with medium or good,
either to avoid being forced into giving an answer that they feel could single them out from the
responding population or just to get the interview over as soon as possible. By keeping an even
rating scale i.e. 1 to 4, the respondents were forced to choose either a negative or positive
response. On the opposite end, rather than keep the rating scale too large, like 1 to 6 or 1 to 10, a
relatively shorter scale saved on time spent to explain what each rating on the scale stood for,
since also the time allocated for each interview was just about 30 minutes per interview.
3.1.3 Analyses of answers received from interviews
While conducting the interviews and collecting the feedback, analyzing the collected data
parallely to develop a system to identify what measurements are useful to evaluate the data
collected from interviews and product usage logs is important. This is done to give a clear idea as
to how operators utilize the WLO and what influences their answers to the questionnaire.
Analysis of data is usually divided into quantitative or qualitative research as shown by
Denscombe (2007). He further shows that:
-
Quantitative research of data is usually associated with analysis
-
Qualitative research of data is usually associated with description.
11
Denscombe (2007) further highlights the five main stages of data analysis in his book as shown
in the table below:
The five main stages of data analysis
1. Data Preparation
2. Initial exploration of the data
Quantitative data
Coding (which normally takes
place before data collection)
Categorising the data
Checking the data
Look for obvious trends or
correlations
3. Analysis of the data
Use of statistical test, e.g.
descriptive statistics, factor
analysis, cluster analysis
Link to research questions or
hypotheses
4. Representation and display of the
data
Tables
Figures
Written interpretation of the
statistical findings
5. Validation of the data
External benchmarks
Internal consistency
Comparison with alternative
explanations
Qualitative data
Transcribing the text
Cataloguing the text or visual data
Preparation of data and loading to
software (if applicable)
Look for obvious recurrent themes
or issues
Add notes to the data
Write memos to capture ideas
Code of the data
Group the codes into categories or
themes
Comparison of categories and
themes
Quest for concepts (or fewer, more
abstract categories) that encapsulate
the categories
Written interpretation of the findings
Illustration of points by quotes and
pictures
Usage of visual models, figures and
tables
Data and method triangulation
Member validation
Comparison with alternative
explanations
Table 1: The five main stages of data analysis (Denscombe, 2007, p.252)
The data dealt with in this thesis is quantitative data as can be clearly seen from table 1.
Denscombe (2007) suggests that it is important for the researcher to be thoroughly familiar with
the data he is dealing with. Eisenhardt (1989) further argues that this familiarity with the data
will allow the researcher to not only identify any pattern in individual cases but also through all
cases together.
In their paper, Gustafsson et al. (2000), draw a link between customer satisfaction and the
product design of Volvo Cars, by suggesting ways in which the customer feedback is linked to
12
the changes in the product design. The authors highlight the need to identify customer needs in
need of improvements as an important step in product design, which will allow the company to
incorporate their feedback in manufacturing products that will satisfy customer requirements.
Using quality function deployment (QFD) and customer satisfaction modeling (CSM), the
authors integrate the two approaches to illustrate the path followed by Volvo Cars from
translating the customer feedback to the means required to accomplish it.
Volvo as a company, have been utilizing QFD since 1988 and have completed over 50 QFD
related projects since then. The usage of this tool has allowed the company to improve their
product development process, a wider knowledge throughout the company regarding Volvo
products, increased involvement from the customer in the development process and more
flexibility to the people involved in the process (Gustafsson et al., 2000). By using their own
customer survey measurement system, Volvo ensures that customers are able to provide both,
tangible and intangible feedback, which the company then uses in the product development
process.
Gustafsson et al. (2000) present an example of this customer survey measurement done by Volvo
with one of their products, the Volvo 850, and show how the customer feedback captured was
utilized in conjunction with actual measurement data from the car, to be able to identify the
problematic areas and subsequently eliminate them. This method of combining customer
preferences and feedback with actual measurement data forms the basis of this thesis, which
ultimately paves the way for improving the quality and the customer satisfaction, Volvo’s
number one priority (Flodin et al., 1997).
3.1.4 Clustering or grouping customers based on their answers
Finally, identifying what data is needed to be able to cluster and group the customers into
categories according to how they value the various product characteristics. The clustering should
be such that Volvo CE can configure individual product variants for each “customer” group. The
results should be discussed with the various stakeholders of the project to see if the method is
value adding to product design. The clustering methodology used is not a single step process as
13
defined by Jain and Dubes (1988). In their paper on Cluster Analysis, they divide the process
into the following steps:
1) Extraction of relevant data from the available data sources which leads to data
collection.
2) Data cleaning/warehousing (Jarke et. al., 1999) or intitial screening of the data after its
extraction.
3) Preparing the data and representing it properly so that it can be clustered.
4) Checking to see if the data has a natural tendency to cluster or not, which is often
ignored in presence of large data sets.
5) Careful choice of clustering techniques and initial parameters.
6) Validating the data, which in their opinion is an under-studied aspect of clustering. They
argue that validation is often based on manual data examination, which is quite
impossible as the amount of data to be clustered grows.
7) Interpretation of the clustering in combination with other studies, to draw conclusions
and further analysis.
Within this thesis work, Volvo CE has taken a decision to first test the method developed by the
thesis author, internally, and upon measuring its success, subsequently then roll out the
questionnaire developed and the method to relate customer feedback to measurements, to its
external customers. Not only will this allow Volvo CE to evaluate the success of the method
developed, it will allow the employees of Volvo CE to contribute to any changes that can be
brought about in the design of the WLO, thus leading to motivation and a sense of being part of
any developments in the new product design. This leads to an establishment of a long term
relationship with the employees, thereby empowering the employees to be innovative (RobertsLombard, 2010; Lucas and Kline, 2008).
14
3.2 Background of tests conducted in September 2011
The measurements conducted in September 2011, covered in the paper by Frank et al. (2012),
were done on three L220F’s of the Volvo CE WLO. The tests were conducted over a 3 week
period in September 2011. External operators as well as internal operators chosen from different
departments at VCE were instructed to perform 3 work tasks at 3 simulated work sites. The 3
applications were:
1. Short Loading Cycle – Gravel
2. Load and Carry Cycle – Gravel
3. Short Loading Cycle – Rock
The operators participating in the measurements were categorized into 4 classes based on their
experience with handling a WLO.
1. External Professionals (EP’s) – Operators who handled/used a WLO as part of their
regular day to day working activities
2. Internal Professionals (IP’s) – Test Operators/Trainers at VCE
3. Internal Averages (IA’s) – Basic knowledge of the WLO
4. Internal Rookies (IR’s) – Participants with 2-10 hrs. WLO operating time
Measurements were taken from each of the operators usage of the WLO’s on the 3 applications.
These measurements were in the form of time data that was collated and analyzed using Matlab.
A camera fitted on the mirror in the WLO cab was used as a validation tool during the analysis.
For this thesis, only the measurements from the internal operators are considered as Volvo CE is
going to evaluate the success of the methodology used and the results obtained before deciding
whether to roll it out to the external operators. Identifying the correlation between how the
operator used the WLO and how they responded to the questionnaire will help to analyze the
gaps occurring in the usage and hence pave the way for better product usage along with the
elimination of problems during the usage or reasons for operator dissatisfaction.
15
3.3 Interview Process
Initially, the questions were derived from previous questionnaires answered by the customers
along with a few added questions from the papers studied in the literature study. A pilot run of
the interviews was performed on selected operators who participated in the measurements in
September 2011, to be able to gauge their responses to the questions and also gather their
feedback as to whether they felt the questions were appropriate enough to cover the basic WLO
characteristics and important parameters. The pilot interview process revealed deficiencies in the
questionnaire as it did not entirely cover the parameters involved in the tests conducted last year.
Once the questions had been reformulated to incorporate the feedback from the test subjects, a
new questionnaire was developed which split the targeted feedback into 2 parts:
1. The tangible questions from which correlations would be investigated to the
measurements
2. The intangible questions which helped the interviewer build a profile of the respondents
based on their experience with driving a WLO.
Once the questionnaire was prepared and finalized, personal interviews were set up with 43 of
the participants. These one-on-one interviews lasted between 30 minutes - 1 hour and comprised
an almost equal distribution of operators from the 3 classes mentioned above. Some of the
questions were answered with a numeric value while some had a rating scale. This was done to
be able to easily codify and store the data rather than having to decipher long answers and extract
the needed information from them.
Overall, the participants seemed very interested in how their answers would influence the
outcome of the study in the thesis. A problem faced was the fact that the previous study
conducted (by Mr. Frank) was more than a year ago and hence most participants struggled to
remember how they handled the machine in September 2011 and to give the answers.
16
4 Analyses of Results
During the operation of the WLO’s in the 3 applications, measurements from signals from the
original CAN-bus in the WLO were captured. The questions in the questionnaire formulated
were based on these signals. Each question is linked to multiple signals and the analyses of these
signals reveal the correlation between the measurements and the answers to the questionnaire.
Since there is no ideal performance measure, the answers to the questionnaire are correlated to
actual usage by the operators in the specified application.
The work cycle in the operation of the WLO in each application is divided into three phases as
below:
1) Bucket Fill – the point from when the bucket gets close enough to the ground up until
when the operator moves the gear shift lever in the reverse direction and leaves the pile.
2) Bucket Empty – approximately from when the operator raises the bucket to a certain
height and moves towards the hauler or load receiver to empty the bucket and then when
the bucket is brought down to a lower height while reversing from the hauler or load
receiver.
3) Transport – Everything else in the phase viz. loading, unloading, forward and reverse.
(Frank, B. et al., 2012) (www.iea.lth.se)
The conditions to be kept in mind are that all operators operated the same WLO, in the same
material, at the same location, only in probably differing weather conditions, in each application.
The analyses in this thesis have been divided into different chapters to bifurcate the types of
questions in the questionnaire and the analysis done for them as below:
Chapter I deals with questions that show how good the operators are at estimating their
performance on the machine
Chapter II deals with questions that allow operators to give ratings and highlight their
preferences on the WLO.
Chapter III deals with the clustering performed, to be able to group operators according to their
answers and their performance.
17
Wheel loader working cycles
Figure 4: Wheel loader working cycles – SLC (Rock & Gravel) and L&C (Gravel)
18
Areas identified as critical areas in a WLO working cycle
Figure 5: Areas identified as critical areas in a WLO working cycle
Figure 5 shows the main areas concentrated upon throughout this thesis, areas identified as
critical areas of interest, from where the operator’s feedback would be very useful: during filling
of the bucket (rock or gravel), transportation to the hauler (in SLC applications) and
transportation over a certain distance (for example, up the hill in L&C applications).
19
Index of questions asked and areas concentrated upon in the report
No.
5.1
5.2
5.3
5.4
5.5
5.6
5.7
5.8
Thesis Question
Operator Class
Engine speed during
IP
bucket fill phase, fuel
IA
efficiently
IR
IP
Vehicle speed during
IA
transport
IR
IP
Cycle Distance
IA
IR
IP
Lifting force during the
IA
bucket fill phase
IR
IP
Engine power
IA
IR
IP
Cycle Times
IA
IR
IP
Fuel Efficiency
IA
IR
IP
Rim pull
IA
IR
SLC (Gravel)
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
-
SLC (Rock)
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
L&C(Gravel)
X
X
X
X
X
X
X
X
X
X
X
X
-
Table 2: Areas of focus during the thesis and classes of operators concentrated upon
After conducting the interviews with the operators, it was decided by the thesis author to shortlist
the above questions in table 2 for analyses. Not all the questions in the questionnaire are thus
included in the analysis, with some being really complex to be able to analyze (the breakout
force), some not having any direct measurements to compare with (question about stability) and
some having no negative respondents (questions about reach and height of boom), that they have
been left out of the analyses. This decision was taken by the thesis author after looking at the
data available from the CAN-bus signals and the answers to the questionnaire by the operators.
20
Chapter I
21
4.1 Engine Speed during bucket fill phase with fuel efficiency
The operators were asked at what engine speed should they operate the WLO during the bucket
fill phase, both, aiming for fuel efficiency and for productivity. Predictably, the engine speed
values while considering the fuel efficiency were lower than those while considering the
productivity, because the operators tended to use more throttle and push the machine harder
while aiming for a higher productivity. As the entire basis of this thesis revolves around the fuel
efficiency of the operators in each application, hence, only the engine speeds the operators
answered while considering the fuel efficiency have been used in the below comparison. A
similar comparison can be made by taking the engine speed answered while considering the
productivity as well.
The engine speeds extracted only for the bucket fill phase for the 3 applications are plotted using
Minitab. All engine speeds below 600 RPM were eliminated to avoid the non-true values, even
though the engine idling speed is about 750 RPM. The median value of the range in which the
operator uses the machine is found. This median value is then compared to the answer given by
the operators to check the difference between how they think the WLO should be utilized and
how they actually use it.
22
4.1.1 Short Loading Cycle (Gravel) and Load & Carry (Gravel)
The graphs below indicate the comparison between the values of the engine speed as answered
by the operators (considering the fuel efficiency) and the actual utilization values (median),
firstly for both the gravel applications.
2000
1800
Engine Speed (RPM)
1600
1400
1200
1000
Answered
800
Median RPM (SLC - Gravel)
600
Median RPM (L&C - Gravel)
400
200
IP16
IP15
IP14
IP13
IP12
IP11
IP10
IP9
IP8
IP7
IP6
IP5
IP3
IP2
IP1
0
Operator ID
Figure 6: Comparison between engine speed values (gravel applications) – IP’s
For the IP class, in figure 6, it can be seen that for most operators, the difference between the
answered values and the actual values is quite low, which indicates a good level of assumption
among the Internal Professionals. The level of estimation is quite high among the IP’s, compared
to the other classes of operators.
23
2000
1800
Engine Speed (RPM)
1600
1400
1200
1000
Answered
800
Median RPM (SLC - Gravel)
600
Median RPM (L&C - Gravel)
400
200
IA18
IA17
IA16
IA15
IA14
IA13
IA12
IA11
IA10
IA8
IA6
IA5
IA4
IA3
IA2
IA1
0
Operator ID
Figure 7: Comparison between engine speed values (gravel applications) – IA’s
Figure 7 shows that 9 out of the 16 operators in the IA class have a good idea about how they
actually use the WLO and how they visualize using it. This is reflected in the low difference
between the median RPM and the answered fuel efficiency RPM. In the case of the operator
IA12, he has overestimated the engine speed at which he has used the WLO whereas for operator
IA16, he has underestimated the engine speed.
24
1800
Engine Speed (RPM)
1600
1400
1200
1000
Answered
800
Median RPM (SLC - Gravel)
600
Median RPM (L&C - Gravel)
400
200
IR14
IR12
IR11
IR10
IR9
IR8
IR7
IR6
IR5
IR3
IR2
IR1
0
Operator ID
Figure 8: Comparison between engine speed values (gravel applications) – IR’s
Lastly, for the IR class, there is more variation between the answers given and the actual engine
speed RPM, as can be seen in figure 8. There are more operators in the IR class who either
overestimate or underestimate the engine speeds at which they operate the WLO.
25
4.1.2 Short Loading Cycle (Rock)
Now, the same data presentation for the SLC (Rock) application is done below.
1800
Engine Speed (RPM)
1600
1400
1200
1000
800
Answered
600
Median RPM (SLC Rock)
400
200
IP16
IP15
IP14
IP13
IP12
IP11
IP10
IP9
IP8
IP7
IP6
IP5
IP3
IP2
IP1
0
Operator ID
Figure 9: Comparison between engine speed values (rock application) – IP’s
For the IP class of operators, in figure 9, we can see lesser difference between the median value
of engine speed and the answers given. This can be attributed to the fact that the IP’s are more
experienced and handle the WLO’s better than the other operators. The standout operators here
are IP12, IP14 and IP16, all of whom have overestimated the engine speed that they used in this
application.
26
2500
Engine Speed (RPM)
2000
1500
Answered
1000
Median RPM (SLC Rock)
500
IA18
IA17
IA16
IA15
IA14
IA13
IA12
IA11
IA10
IA8
IA6
IA5
IA4
IA3
IA2
IA1
0
Operator ID
Figure 10: Comparison between engine speed values (rock application) – IA’s
Looking at the graphs for SLC (Rock), we find more variation between the answers by the
operators and the actual usage of the WLO. The SLC application for rock handling is harder and
requires more skill than the SLC application for gravel handling. This is reflected in figure 10,
for the IA’s, where there are more operators who have overestimated the engine speed they
answered for the application in comparison to their actual usage.
27
1800
Engine Speed (RPM)
1600
1400
1200
1000
800
Answered
600
Median RPM (SLC Rock)
400
200
IR14
IR12
IR11
IR10
IR9
IR8
IR7
IR6
IR5
IR3
IR2
IR1
0
Operator ID
Figure 11: Comparison between engine speed values (rock application) – IR’s
Lastly, in the IR class of operators, in figure 11, a much larger difference can be observed. The
standout operator here is IR1, in whose case, a 902 RPM difference can be observed. The large
variation for the rookie operators can be attributed to their inexperience on the WLO’s, as for
many of them, the tests conducted for Mr. Frank’s project, was their first time operating the
WLO in the rock application.
28
The input signals associated with the engine speed (RPM) is the accelerator pedal position (usage
in percentage). Taking a detailed look at the accelerator pedal position reveals an obvious trend
that was assumed beforehand: there exists a positive correlation between the two signals. An
increase in the usage of the accelerator pedal results in an increase in the engine speed.
Conversely, an increase in engine speed is because of an increase in the usage of the accelerator
pedal.
Now, the analysis of the accelerator pedal position for all operators only shows the following:
1) How much of the accelerator pedal does one operator use in comparison to another
2) How the engine speed increases as an operator uses more of the accelerator pedal (as
stated before).
The below graph shows how much of the bucket fill phase does each operator spend by pushing
full throttle on the WLO. This can be a good indication of how each operator utilizes the full
throttle available to them during this phase. The graph shows that some operators spend more
than 50% of the time in the bucket fill phase using the full throttle. An obvious correlation, as
stated before, is that the more of the throttle you use, the higher speed you get from the engine.
This could be one of the reasons why some operators answer a higher RPM in the questionnaire
than others, because they use more of the throttle compared to others.
29
100% Accelerator Pedal Usage (% in bucket fill phase)
80
70
60
50
40
%
30
20
10
IP1
IP2
IP3
IP5
IP6
IP7
IP8
IP9
IP10
IP11
IP12
IP13
IP14
IP15
IP16
IA1
IA2
IA3
IA4
IA5
IA6
IA8
IA10
IA11
IA12
IA13
IA14
IA15
IA16
IA17
IA18
IR1
IR2
IR3
IR5
IR6
IR7
IR8
IR9
IR10
IR11
IR12
IR14
0
Operator ID
Figure 12: % of bucket fill phase spent using 100% accelerator pedal – SLC (Gravel)
As can be observed in figure 12, there is no conceivable pattern as to how operators use the full
throttle i.e. whether the IP’s utilize lesser due to better machine control or whether the IR’s use
more due to lesser experience on the machine.
Another analysis of a one to one comparison can also be done, by comparing how much
percentage of the bucket fill phase does an operator utilize different accelerator pedal position
percentages in comparison to the ideal operator (in this case, the ideal operator is the most fuel
efficient operator in the respective application). To show how this works, a comparison between
IA1 and IP2 (the most fuel efficient in SLC - Gravel) is done to show how much percentage of
the bucket fill phase these operators spend in each bin of 10% accelerator pedal usage. As can be
seen from the below graph, both these operators utilize a quite high percentage of the full throttle
during the bucket fill phase. Also, the operator IP2 concentrates on using more of the full throttle
than operator IA1 during this phase. As can be seen, IP2 relies on using the full throttle 57% of
the time spent in the bucket fill phase, which gives an indication of a high engine speed RPM
value.
30
70
60
% of usage
50
40
IP2
30
IA1
20
10
0
~ 10
~ 20
~ 30
~ 40
~ 50
~ 60
~ 70
~ 80
~ 90
~ 100
% range of accelerator pedal usage
Figure 13: Comparison between % usage of accelerator pedal position for IP2 and IA1 – SLC
(Gravel)
This method of comparison, as in figure 13, can be done for all the operators, either against the
most fuel efficient operator or within themselves. Further, this can also be done for the other
applications to highlight the difference in how each operator uses the accelerator pedal during the
bucket fill phase. This presents an idea about which operator uses the full throttle more compared
to the others and how it differs in each application.
31
Conclusions
This difference, in the opinion of the thesis writer, can be attributed to the fact that the IP
operators are more experienced in handling a WLO compared to the other two classes of
operators and hence have a more accurate estimation of the operating engine speed. This also
explains the low variation between the three compared values.
The larger variation among the three values in the IA and IR class, again in the opinion of the
thesis writer, can be attributed to the following:
-
Inexperience in handling the machine.
-
Feeling of a stressed environment; in the SLC (Gravel) application, the constricted space
coupled with constantly having to avoid hitting the articulated hauler; in the L&C
(Gravel) application, the need to fill the bucket as much as possible due to the travel
involved.
What is also noticeable is that while most IP operators rely on their experience to handle the
WLO at more medium speeds, the IA’s and the IR’s tend to push the machine to higher RPM’s
to perform the applications, due to their inexperience in driving a WLO.
32
4.2 Vehicle Speed during transport
The operators were asked at what maximum speed they should drive the WLO during the
transportation phase of the applications. This gives an idea of the maximum speed at which the
operators feel most ideal to drive the WLO at i.e. at what maximum speed they would like the
machine to perform or be driven at. After mining the data from the signals, the transportation
phase in the three applications were chosen for analysis. The below graphs indicate the
difference in actual maximum speed at which the WLO was driven during the tests and the speed
which the operators answer that the WLO should be driven at.
33
4.2.1 Short Loading Cycles (Gravel & Rock)
14
Vehicle Speed (km/hr)
12
10
8
Answered
6
SLC (Gravel)
4
SLC (Rock)
2
IP16
IP15
IP14
IP13
IP12
IP11
IP10
IP9
IP8
IP7
IP6
IP5
IP3
IP2
IP1
0
Operator ID
Figure 14: Comparison between maximum vehicle speeds for SLC applications – IP’s
In the case of the IP class of operators, 3 out of the 15 operators have a larger difference in the
answered value and the actual values. As can be seen from figure 14, the operators tend to
answer a value much lower than what they actually drive at. A similar trend follows for the other
two classes of operators as well as can be seen from the below graphs.
34
16
Vehicle Speed (km/hr)
14
12
10
8
Answered
6
SLC (Gravel)
4
SLC (Rock)
2
IA18
IA17
IA16
IA15
IA14
IA13
IA12
IA11
IA10
IA8
IA6
IA5
IA4
IA3
IA2
IA1
0
Operator ID
Figure 15: Comparison between maximum vehicle speeds for SLC applications – IA’s
The SLC applications with very less distance travelled during the working phases cause the
WLO to not reach very high speeds. 4 out of the 16 operators have underestimated the speeds at
which they actually drive the machine and this is reflected in the difference of more than 5
km/hr. in their answers, as shown in figure 15 above.
35
18
Vehicle Speed (km/hr)
16
14
12
10
Answered
8
SLC (Gravel)
6
SLC (Rock)
4
2
IR14
IR12
IR11
IR10
IR9
IR8
IR7
IR6
IR5
IR3
IR2
IR1
0
Operator ID
Figure 16: Comparison between maximum vehicle speeds for SLC applications – IR’s
1 out of the 12 operators in the IR class has a large difference in the vehicle speed values, as
shown above in figure 16 and this is the largest difference of value among the 3 classes of
operators – 9 km/hr. for operator IR10.
36
To further investigate why this difference occurs, the associated signal with the vehicle speed
was looked at: the actual gear utilized during this phase of the application. The top 5 most fuel
efficient operators were chosen to see if there is any trend developing. The noticeable
development from analyzing the vehicle speeds and the actual gear at these speeds for these
operators was that the maximum speeds were achieved at a negative gear i.e. when the operators
were reversing the vehicle, not while going forward. The forward transportation is much more
constricted as the operators have the gravel pile in front of them in one forward motion and the
articulated hauler in the other forward motion. This constraint is however not there when
reversing, as the operators can reverse faster in both the backward motions, from the pile and
from the hauler. This is found to be true in both the SLC applications for Gravel and Rock.
Conclusions
One conclusion that can be drawn from this is that when the operators answered the question,
they focused only on the forward movement than the reversing motion which explains the larger
discrepancy in their answers.
Also, the feedback received from the operators was that rather than concentrating on what speed
they drove the vehicle at, they were more concerned about the bucket fill factor and hence did
not pay attention to what speed the vehicle was driven at. To get a better idea about the vehicle
speeds, it needs to be explicitly stated to them to note it for future tests.
37
4.2.2 Load & Carry (Gravel)
The third application chosen for analysis was the Load & Carry cycle for Gravel handling
application. This application involved the operators filling up the bucket of the WLO with gravel
and driving up a hill to deposit the material into a hopper. This allowed for the WLO to be driven
at a higher speed due to longer transportation involved.
35
Vehicle Speed (km/hr)
30
25
20
15
Answered
10
Achieved
5
IP16
IP15
IP14
IP13
IP12
IP11
IP10
IP9
IP8
IP7
IP6
IP5
IP3
IP2
IP1
0
Operator ID
Figure 17: Comparison between maximum speed values for L&C (Gravel) – IP’s
In the IP class of operators, 5 operators have a difference of more than 5 km/hr. between the
maximum values they drove the vehicle at and their answers. As can be observed from the figure
17 above, these operators have underestimated the vehicle speeds at which they drive the WLO.
This can be attributed to the fact that these operators have used accelerator pedal a lot more than
the other operators to be able to get the vehicle up to higher speeds while driving.
38
40
Vehicle Speed (km/hr)
35
30
25
20
Answered
15
Achieved
10
5
IA18
IA17
IA16
IA15
IA14
IA13
IA12
IA11
IA10
IA8
IA6
IA5
IA4
IA3
IA2
IA1
0
Operator ID
Figure 18: Comparison between maximum speed values for L&C (Gravel) – IA’s
In the IA class of operators, 5 out of the 16 operators have a difference of more than 5 km/hr.
between the maximum value of the speed at which they have driven the vehicle and the answer
as to what maximum speed they should drive the vehicle, the standout operator being IA17 with
a 14 km/hr. speed difference (see figure 18).
39
35
Vehicle Speed (km/hr)
30
25
20
Answered
15
Achieved
10
5
IR14
IR12
IR11
IR10
IR9
IR8
IR7
IR6
IR5
IR3
IR2
IR1
0
Operator ID
Figure 19: Comparison between maximum speed values for L&C (Gravel) – IR’s
Lastly, in the IR class, 2 of the operators have a significant difference of more than 10 km/hr.
between the max speed and their answers. The IR class of operators being inexperienced, as
compared to the IA and IP class, show more variation in their answers, either over or
underestimating the vehicle speed values (see figure 19).
40
Conclusions
The L&C (Gravel) application involved more forward motion than reversing motion. As can be
observed from figures 17, 18 and 19, the level of variation is a lot lesser than in the previous
applications of the SLC. The discrepancy observed in this application can be attributed to the fact
that while the operators answered the maximum speed achievable in a L&C application was over
a flat surface with a distance of 100 m (approximately), the actual length of the course in the
L&C test was about 120 m (approx.) as shown in figure 20 below and a part of the course was a
hill, up which the WLO had to be driven.
Figure 20: Approximate location of the L&C course in the test
Due to this, it can be seen that while some operators have a good estimate of the maximum
speeds (achievable and achieved); most of them have some variation. The response most
received from the operators was that they did not notice the speed at which they drove and that
their answers were just assumptions, because the vehicle speed was not an important factor while
driving as compared to the engine speed or bucket fill factor to them.
41
4.3 Cycle Distance
The cycle distance is the distance covered by the WLO during the operation of a single cycle in
an application. A total of 15 cycles were performed by each operator. For this analysis, the
operators were asked how long of a distance did they cover in a single cycle during the SLC
applications. The answers given by them are then compared to the median value obtained from
the below boxplots to identify whether they over or underestimate the distance required to
complete one cycle.
42
IA1
IA2
IA3
IA4
IA5
IA6
IA8
IA 10
IA 11
IA 12
IA 13
IA 14
IA 15
IA 16
IA 17
IA 18
IP1
IP2
IP3
IP5
IP6
IP7
IP8
IP9
IP10
IP11
IP12
IP13
IP14
IP15
IP16
IR1
IR2
IR3
IR5
IR6
IR7
IR8
IR9
IR10
IR11
IR12
IR14
Cycle Distance (m)
IA 1
IA 2
IA 3
IA 4
IA 5
IA 6
IA 8
IA 10
IA 11
IA 12
IA 13
IA 14
IA 15
IA 16
IA 17
IA 18
IP1
IP2
IP3
IP5
IP6
IP7
IP8
IP9
IP10
IP11
IP12
IP13
IP14
IP15
IP16
IR1
IR2
IR3
IR5
IR6
IR7
IR8
IR9
IR10
IR11
IR12
IR14
Cycle Distance (m)
Boxplot of Cycle Distance - Taltet
70
60
50
40
30
20
Figure 21: Boxplot of Cycle Distances – SLC (Gravel)
Boxplot of Cycle Distance - Berg
120
100
80
60
40
20
0
Figure 22: Boxplot of Cycle Distances – SLC (Rock)
43
100
90
Cycle Distance (m)
80
70
60
50
Answered
40
Taltet Median
30
Berg Median
20
10
IP16
IP15
IP14
IP13
IP12
IP11
IP10
IP9
IP8
IP7
IP6
IP5
IP3
IP2
IP1
0
Operator ID
Figure 23: Cycle Distance comparison for IP
In the analysis for the IP class of operators, the operators have a better idea of the cycle distance
they covered, which is reflected in lesser variation between their answers and the measured
values. Here too, it can be observed that for most of the operators, their answers are lesser than
both the SLC application median values, as can be seen in figure 23 above. The one standout
operator, IP15, has a larger difference in value, though this is not indicative of the general
opinion of the IP class of operators. An assumption that the said operator misunderstood the
question could explain the large value as said by him.
44
120
Cycle Distance (m)
100
80
60
Answered
Taltet Median
40
Berg Median
20
IA18
IA17
IA16
IA15
IA14
IA13
IA12
IA11
IA10
IA8
IA6
IA5
IA4
IA3
IA2
IA1
0
Operator ID
Figure 24: Cycle Distance comparison for IA
As can be observed from figure 24, most of the operators in the IA class have underestimated the
cycle distance they covered to complete a cycle. This is reflected in their answers being lesser
than the measured values. It can also be seen that these operators have underestimated their
answers in relation to both the SLC (Gravel) and SLC (Rock) application. However, two
operators, IA4 and IA5 have overestimated their answers by a large value.
45
100
90
Cycle Distance (m)
80
70
60
50
Answered
40
Taltet Median
30
Berg Median
20
10
IR14
IR12
IR11
IR10
IR9
IR8
IR7
IR6
IR5
IR3
IR2
IR1
0
Operator ID
Figure 25: Cycle Distance comparison for IR
Lastly, in the IR class, it can be observed from figure 25 that the variation is more uneven, with
more operators answering with values of the cycle distance greater than or equal to the SLC
(Rock) application. The operators IR8 and IR9 have overestimated, whereas operator IR10 has
underestimated the value of the cycle distance.
The analysis for cycle distance reveals that in most cases, the operators underestimate the
distance they covered to complete a cycle in either of the applications. Though, there are a few
standout points in each class of operators, it can be concluded that most of the operators drive the
WLO for a longer distance than they think they do.
46
Figure 26 below, shows the average cycle distance covered by each operator against the average
cycle time taken by each operator to complete the cycle, in the SLC (Gravel) application (Here,
the average and the median values are the same and hence the average was chosen). It can be
observed that the IA and IP operators are clustered quite close together. Due to the experience
possessed by the IP’s, not only do they take lesser time to complete a cycle, but they also do so
in a shorter distance compared to the other two classes. Quite predictably, the IR operators take
longer, both in time and the distance to complete the task.
80
Average Cycle Time (s)
70
60
50
40
IP
30
IA
IR
20
10
0
0
10
20
30
40
50
60
Average Cycle Distance (m)
Figure 26: Average Cycle Distance vs. Average Cycle Time – SLC (Gravel)
47
120
Average Cycle Time (s)
100
80
IP
60
IA
40
IR
20
0
0
10
20
30
40
50
60
70
80
90
100
Average Cycle Distance (m)
Figure 27: Average Cycle Distance vs. Average Cycle Time – SLC (Rock)
Figure 27 is the average cycle distance vs. the average cycle time for each operator in the SLC
(Rock) application. As before, it can be observed that the IP operators take lesser time and
shorter distance to complete the task compared to the IA and IR classes of operators.
48
Conclusions
This could be one of the reasons that though both these applications are short loading cycles, the
operators cover more distance in the rock task compared to the gravel task. This could also be the
reason that the cycle time for the rock application for all operators is longer than the cycle time
in the gravel application.
Also, as can be seen from this section, the IP operators cover the least distance while performing
the SLC applications and do so in the fastest time, while the IR operators cover the most distance
and take the longest time. The conclusions drawn from this analysis are as stated below:
1) The IP operators perform the applications the fastest as compared to the other two classes
of operators, due to a higher skill level.
2) The SLC (Rock) application takes longer to complete than the SLC (Gravel) application
because:
- It is an application that requires more skill.
- The operators need to plan which part of the material pile they fill the bucket with due
to the larger material in the rock application.
3) The distance travelled in the L&C application is much more than the other two
applications.
49
Chapter II
50
This chapter of the report deals with how the operators would prefer the machine to operate.
Their preferences are captured using a rating scale, which shows that those operators that are
satisfied with the machine performance give a higher rating and those that want the machine to
perform better, give a lower rating. The rating scale is as below:
1 – Poor
2 – Not good enough
3 – Nearly good enough
4 – Good enough
4.4 Lifting force during the bucket fill phase
The lifting force during the bucket fill phase is derived from the Load sensing pressure
(LsPressure). Depending on whether the levers for lifting and/or tilting are being used, the
LsPressure is defined by the force required to fill the bucket. The operators were asked whether
they felt that the lifting force was enough while they filled the bucket and their answers were
captured on the rating scale.
The application chosen for this analysis was SLC (Rock) as there were no negative responses in
the gravel applications.
In the SLC Rock application, 8 out of the 43 operators answered that they felt that the lifting
force was not enough. The maximum pressure imparted from the 1st pump in the WLO is 24
MPa (± 0.5 MPa) and from the 2nd pump is 26 MPa (± 0.5 MPa) (Hydraulic system
specifications for L220F – Operator’s Manual for L150F/L180F/L220F). The lifting force or the
LsPressure comes from the 1st displacement pump and if more is required, the 2nd displacement
pump imparts the required pressure. The presence of a “priority spool” on the 2nd pump decides
if the pressure should go to the steering of the WLO or towards the bucket to help fill it up, but
during the bucket fill phase, the steering is not used and hence, this is neglected. The usage of the
levers during the bucket fill phase in relation to what pressure it indicates is given in table 3.
51
This is a simplified version of the lever used and the associated LsPressure with it. There is a
more detailed explanation of this association, but it is not covered in the report as it does not
come under the scope of this thesis.
Lever
LsPressure
No lever
Standby Pressure
1st
Lifting force
2nd
Tilting force
1st + 2nd
Tilting force
Table 3: Lever used and associated pressure choice
When the 1st and the 2nd levers are being used together to lift and tilt the bucket to fill it up, the
WLO senses which of the two pressures – Lifting or Tilting, is higher and assigns the LsPressure
as the one which is the lower among the two, which is mostly, the Tilting pressure.
To analyze why some operators respond negatively and some positively, a deeper analysis was
required on the LsPressure and the factors affecting it. The bucket fill phase in the SLC (Rock)
application was chosen for this analysis. Firstly, to ensure that the LsPressure values measured
were in fact the lift force values, only those values wherein the 2nd lever position was set to zero
and the 1st lever position was positive, were considered. The bucket fill phase is best carried out
when the 1st gear is engaged on the machine and hence only those LsPressure values when the
machine was in gear 1 was filtered.
The most fuel efficient operator in this application, IP2, was chosen as a benchmark to compare
the dissatisfied operators against. This operator is also the most fuel efficient in his class of
operators, while operators IA8 and IR9 are the most fuel efficient operators among the IA’s and
IR’s respectively, in the SLC (Rock) application.
The first noticeable difference between operator IP2 and the remaining dissatisfied operators is
that IP2 spend almost the entire bucket fill time using the full throttle i.e. 100% of the accelerator
pedal position. While some of the dissatisfied operators use a high percentage of the accelerator
pedal as well, none of them are even close to operator IP2 in this aspect.
52
Operator 1st Lever Position LsPressure Output Torque Engine Speed Vehicle Speed Accelerator Pedal Position
IP2
127.00
27.30
1545.57
1636.39
0.82
100.00
IP1
75.00
26.18
1437.20
1748.00
0.10
100.00
IP14
48.00
28.52
1627.02
1596.36
1.84
85.20
IP15
12.00
27.50
1491.39
1663.36
1.54
80.80
IA3
44.00
27.74
1437.20
1677.64
1.23
100.00
IA5
78.00
26.91
1382.90
1646.89
0.20
100.00
IA8
126.00
23.64
1762.53
1370.36
1.64
78.40
IA16
114.00
27.79
1437.20
1621.27
0.82
100.00
IA18
22.00
25.88
1410.05
1579.76
0.20
100.00
Table 4: Comparison of signals at maximum used LsPressure - SLC (Rock)
As can be seen from table 4, the maximum utilized Lift force values for these operators are very
close to each other. What differ are the vehicle speeds at which these operators use the lift force.
Though there is no big difference between the vehicle speeds, is it enough to cause the
dissatisfied operators to respond negatively? Or do they feel the lack of lifting force due to the
fact that they utilized lesser output torque from the engine? Or do some operators achieve this
“maximum” value at a lower lifting height and feel they do not have enough lifting force as they
raise the bucket more?
The presentation of the results in the above table does not answer these questions nor are they
enough to arrive at a conclusion as to why certain operators responded negatively to the question.
What can also be analyzed is the average load carried by these operators in the bucket to see is
there any relation between more load in the bucket and a feeling of lack of lifting force.
The following figures 28, 29 and 30 present the maximum lift force (LsPressure) utilized by the
operators versus the vehicle speeds, accelerator pedal position and the 1st lever positions at which
the lift forces were achieved. This is done to see if any clusters or patterns emerge from the data,
for example, whether certain classes of operators achieve certain lift force values at particular
vehicle speeds and so forth.
53
35
LsPressure (MPa)
30
25
20
IP
15
IA
10
IR
5
0
0
0.5
1
1.5
2
2.5
Vehicle Speed (km/hr)
Figure 28: Achieved LsPressure vs. vehicle speeds for all operators – SLC (Rock)
35
LsPressure (MPa)
30
25
20
IP
15
IA
10
IR
5
0
0
20
40
60
80
100
120
Accelerator Pedal Position (%)
Figure 29: Achieved LsPressure vs. accelerator pedal position for all operators – SLC (Rock)
54
35
LsPressure (MPa)
30
25
20
IP
15
IA
10
IR
5
0
0
20
40
60
80
100
120
140
1st Lever Position (%)
Figure 30: Achieved LsPressure vs. 1st lever position for all operators – SLC (Rock)
Conclusions
As can be seen from the above graphs, no clear clusters or patterns emerge from the usage data.
No clear conclusions can thus be drawn.
55
4.5 Engine Power
The power of the engine is reflected in its ability to support the three important phases during the
operation of the WLO, the bucket fill phase, the transportation phase and the point where the
operator reverses the WLO and drives the WLO forward with load in the bucket to dump it into
the hauler. To analyze the engine power, as an example, the bucket fill phase for the three
applications was chosen. Also, the transportation phase in the L&C application was chosen to
analyze the power, as this phase in the two SLC applications is very less.
The engine power is not a signal used during the tests conducted. But, it is possible to calculate
the power using the formula as given below:
= ∗
2
∗ /1000
60
The engine speed and output torque values are found in the signals used in the tests. 5 of the 43
operators interviewed were of the opinion that the power of the engine was not sufficient. To
further analyze this, the bucket fill phase for the applications was chosen, as an example.
Now, what is more important to the performance of the WLO? The engine power or the torque
exerted by the ICE? The power of the engine is a derived value; it depends on the torque of the
engine.
In simple terms, the torque of the engine is what is needed to perform the work and the power of
the engine is what makes it accelerate and go faster. The maximum torque utilized by all the
operators in the bucket fill phase is the maximum torque allowed by the machine; hence no
conclusions can be drawn from it. Therefore, the median value of the torque over the entire phase
is considered to give a better idea of usage.
56
Out of the 43 operators interviewed, 4 operators answered that they were not satisfied with the
engine power. 3 of the operators belong to the IP class and 1 operator belongs to the IA class.
Rather than look at the speeds or the power in the bucket fill phase, the output torque was looked
at. The below graphs show the median value of the output torque utilized by the operators in the
Median Output Torques (N-m)
three applications during the bucket fill phase.
Taltet Median
Bandet Median
IP16
IP15
IP14
IP13
IP12
IP11
IP10
IP9
IP8
IP7
IP6
IP5
IP3
IP2
IP1
Berg Median
Operator ID
Figure 31: Comparison between median values of output torque – IP’s
The 3 dissatisfied operators in this class, IP8, IP12 and IP16 are presented on figure 31. As can
be seen from the figure, their median values are not as high as compared to some of the other
operators. Though, these operators have at some point in the cycle, used the highest available
torque in the machine, they have operated the WLO at a lower output torque value as compared
to the most fuel efficient operator in this class, IP2.
57
Average Bucket Loads (ton)
Taltet
Bandet
IP16
IP15
IP14
IP13
IP12
IP11
IP10
IP9
IP8
IP7
IP6
IP5
IP3
IP2
IP1
Berg
Operator ID
Figure 32: Comparison of average bucket loads – IP’s
Also, taking a look at the average loads filled in the bucket by these operators in figure 32, these
3 operators have not filled the bucket as much as IP2. This presents a scenario of filling the
bucket to a lesser extent, using lesser of the output torque as compared to the most fuel efficient
operator, but still feeling a lack of power in the machine. One of the conclusions that can be
drawn from this is that the operator did not utilize the full machine capacity, and still complained
about the lack of power.
58
The same analysis for the IA class of operators is done below in figures 33 and 34, to reveal the
same conclusion. The dissatisfied operator, IA4, operated the WLO at a lower value of the output
torque as compared to the other operators in his class, though this operator has filled the bucket
Median Output Torques (N-m)
more than the other IA’s.
Taltet Median
Bandet Median
IA18
IA17
IA16
IA15
IA14
IA13
IA12
IA11
IA10
IA8
IA6
IA5
IA4
IA3
IA2
IA1
Berg Median
Operator ID
Average Bucket Loads (ton)
Figure 33: Comparison between median values of output torque – IA’s
Taltet
Bandet
IA18
IA17
IA16
IA15
IA14
IA13
IA12
IA11
IA10
IA8
IA6
IA5
IA4
IA3
IA2
IA1
Berg
Operator ID
Figure 34: Comparison of average bucket loads – IA’s
59
A similar data presentation for the IR’s can also be done, though none of the IR class of
operators were dissatisfied with the engine power.
Conclusions
-
The IP class of operators operates the WLO at a higher range value of the output torque
while ensuring a higher fill factor in the bucket, compared to the IA and IR class of
operators.
-
Operating the WLO at a lower range value of the output torque causes some of the
operators to feel that the machine does not possess enough power. This can be attributed
to different ways that different operators drive and handle the machine.
-
The problem does not lie with the machine, rather with the operator and the way he
handles the WLO. Overfilling the bucket and using a lower median value of the output
torque can contribute to the operator’s feeling of a lack of power with the machine.
60
Gradeability (power while climbing up a hill)
The Load & Carry cycle for the Gravel application was one of the tests conducted for the
operators. It involved the transport of material from the pile, up a hill, into a hopper. The bucket
fill phase of this application is similar to the phase in the SLC for Gravel, but it involved a
transportation phase which allowed the operators to drive the machine longer and test the power
of the engine up the hill. The operators were asked whether they felt that the WLO had enough
power while climbing the hill. The ratings they gave were then compared to the measurements to
identify any gaps and cluster the operators based on their ratings.
The challenge during the analysis of this question was to identify only the hill that was climbed,
during the transportation phase. This was made difficult by the fact that though the latitude and
longitude coordinate signals for the hill remained the same, every operator did not start his ascent
up the hill nor end the climb at the same point. This difficulty was overcome by cross verifying
the climb up the hill for every operator, for each cycle in the application.
The application involves driving up the hill so that when the operator reached the hopper, all he
has to do is tilt the bucket downwards and empty the gravel into a hopper. 17 of the 43 operators
interviewed felt that the WLO did not have enough power to do so, as the pumps in the machine
withdrew power from the driving and supplied it into the lifting of the boom. This was explained
as one of the reasons that the machine slowed down and went from a higher gear to a lower one,
thereby causing the operators to have to push the accelerator pedal more to be able to reach the
hill top.
61
The presentation of the data for the IP operators revealed a more alarming result, with 10 of the
15 IP operators dissatisfied with the engine power, as can be seen in figure 35. Though the
average output torque utilized by these operators was lesser than the IA operators, the overall
consensus was more of dissatisfaction compared to the IA’s. The IP operators are more
experienced than the other class of operators and it can be safely assumed that any problems such
as lack of power are more easily recognized by the IP’s than the IA’s and the IR’s. While it can
be easily said that those operators who did not utilize as much of the torque as was available
would feel that the machine lacked power, a deeper analysis was required to be able to truly
IP16
IP15
IP14
IP13
IP12
IP11
IP10
IP9
IP8
IP7
IP6
IP5
IP3
IP2
IP1
Output Torque (N-m)
determine the cause of such answers.
Operator ID
Figure 35: Average output torque utilized while climbing the hill – IP’s
62
Figure 36 below, shows the average output torque utilized over all the cycles for each IA
operator during the hill climbing phase in the L&C application for Gravel. The red boxes
indicate the operators who are dissatisfied with the power of the engine during this phase. The
reasons for this could be the fact that the machine truly lacks power while climbing a hill or the
IA18
IA17
IA16
IA15
IA14
IA13
IA12
IA11
IA10
IA8
IA6
IA5
IA4
IA3
IA2
IA1
Output Torque (N-m)
satisfied operators were not able to recognize the lack of power in the machine.
Operator ID
Figure 36: Average output torque utilized while climbing the hill – IA’s
63
Lastly, in the case of the IR operators, only 2 of the 12 operators interviewed felt the lack of
power in the WLO (see figure 37). It is interesting to note that neither of these operators has
utilized as much output torque as some of the other operators while climbing the hill. This can be
attributed to the lack of experience for the IR operators who have not spent more than 2 ~ 10
IR14
IR12
IR11
IR10
IR9
IR8
IR7
IR6
IR5
IR3
IR2
IR1
Output Torque (N-m)
hours on driving the WLO.
Operator ID
Figure 37: Average output torque utilized while climbing the hill – IR’s
64
To truly analyze the answers to this question, the most fuel efficient operator in this application,
IP1, was chosen. To determine if this was a genuine problem, the average output torque in each
cycle, while climbing up the hill, was looked at. Figure 38 shows how the average (median and
Output Torque (N-m)
average same in this case) output torque varies in each cycle for this operator.
Average O.T
1
2
3
4
5
6
7
8
9
10
11
Cycle Number
Figure 38: Average output torque used in each cycle while climbing the hill – IP1
As can be seen from the figure, the average output torque is less in the beginning of the
application as the operator gets used to driving the machine up the hill and increases with the last
3 cycles when the operator has figured out how to drive the machine up the hill better.
This still does not present the answer as to why this operator is dis-satisfied with the engine
power while climbing the hill. To further unravel the reason, one of the cycles was chosen to see
how the output torque, engine speed and gear utilized, varies over the time (time here being the
time spent driving up the hill).
65
Variation of engine
speed in cycle 1
2500
1500
1000
Output
Torque
500
Engine Speed (RPM)
2000
1500
1000
500
0
Time (ms)
Time (ms)
Variation of gear in
cycle 1
Variation of engine
power in cycle 1
3.00
2.00
Gear
1.00
1
18
35
52
69
86
103
120
0.00
Time (ms)
350
300
250
200
150
100
50
0
Power
1
18
35
52
69
86
103
120
4.00
Engine Power (kW)
5.00
Gear Used
Engine
Speed
1
18
35
52
69
86
103
120
0
2000
1
21
41
61
81
101
121
Output Torque (N-m)
Variation of output
torque in cycle 1
Time (ms)
Figures 39: Variation of signals over time in cycle 1 for operator IP1 while climbing the hill
As can be seen from figure 39, an example of cycle 1 was chosen to see how certain signals vary
over time (in ms). Though the output torque initially is low as the operator starts the ascent up
the hill, the torque starts to increase initially as the machine automatically shifts the gear down
from gear 4 to 3 to maintain a higher output torque. Even as the gear remains at 3, the output
torque remains high for a few milliseconds before dropping again. At this point, the operator
pushes the accelerator pedal down to increase the engine speed. The output torque again
increases for a few milliseconds before reducing again. At this point, the machine again shifts a
gear down from 3 to 2 to maintain the higher output torque. This lasts for a few milliseconds
66
more before the torque value reduces again. To avoid the machine coming to a complete
standstill, the operator pushes the accelerator pedal again to raise the engine speed and push the
machine over the top of the hill.
This is a trend that follows in almost every cycle for every operator. While some operators tried
to circumvent this by driving the machine on a lower gear, in that scenario, the output torque
increased at first, but then reduced as the machine went up the hill, on a constant gear. This
analysis can be performed on the other operators to notice the same trend for all the operators.
The most received answer for this question was that the WLO did not possess enough power
while going up the hill, as when it was driven on automatic gears, the WLO kept gearing down
from a higher to a lower gear, thus causing the WLO to slow down while going up the hill.
From the analysis for this question, it can be observed that the IP operators are the ones most
dissatisfied with the engine power while climbing the hill, with a third of the IA’s voicing the
same concerns. This can be interpreted as a real problem area in the WLO because even if we
discount the IR operators, nearly 50% of the IA and IP operators feel that the engine does not
possess enough power during the hill climbing phase in this application and this could be
attributed as a weak area in the WLO as the engine is unable to support driving up the hill.
Conclusions
The analysis of this area of operation reveals the following:
-
The problem with the lack of power in the WLO may really be a problem with possibly
the gear shift time i.e. the time required for the machine to shift from one gear to another.
To ensure this, a deeper analysis is required with the gear shift time parameter.
-
The operator quite possibly lifts the boom as he drives the WLO up the hill, contributing
to lesser power available to drive the machine, which could be a reason that the operator
feels that the machine lacks power.
67
4.6 Cycle Times
The cycle time is defined as the time of one full cycle of operation in any of the applications. The
operators were asked if they measured the cycle time while operating the WLO, but none of the
interviewed operators had done so. They were of the opinion that concentrating on filling the
bucket and maneuvering the WLO to load the material onto an articulated hauler or into the
hopper was more of a priority than measuring the cycle time. The operators also did not have
definitive answers as to what they thought the cycle time would be during an application. Hence,
it is not possible to have a comparison between their answers and their actual cycle times.
Nevertheless, the below box plots highlight the cycle times of each operator in each of the 3
applications.
68
4.6.1 Cycle Times for Short Loading Cycle (Gravel)
Boxplot of Cycle Times - Taltet
110
100
90
Time (s)
80
70
60
50
40
30
IA 1
IA 2
IA 3
IA 4
IA 5
IA 6
IA 8
IA 10
IA 11
IA 12
IA 13
IA 14
IA 15
IA 16
IA 17
IA 18
IP1
IP2
IP3
IP5
IP6
IP7
IP8
IP9
IP10
IP11
IP12
IP13
IP14
IP15
IP16
IR1
IR2
IR3
IR5
IR6
IR7
IR8
IR9
IR10
IR11
IR12
IR14
20
Figure 40: Boxplot of cycle Times for all operators – SLC (Gravel)
As can be seen from figure 40, operators IP2 is the fastest performing operator with the least
cycle time compared to the other operators. Also, the variation within each class of operators is
lesser among the IA and IP class of operators, as compared to the variation among the IR
operators. This can be attributed to the experience of the IA and IP operators and the higher skill
they possess in operating the WLO compared to the IR operators.
Since the operators were not able to quantify the cycle time, a deeper analysis was required. By
splitting the cycle time into its components, time spent in each phase; bucket fill, bucket empty
and transport, a comparison could be made into the time spent by each operator while filling the
bucket versus the time spent in standing still (bucket empty).
69
40
35
Time (s)
30
25
20
% Bucket Fill Time
15
% Bucket Empty Time
10
5
IP16
IP15
IP14
IP13
IP12
IP11
IP10
IP9
IP8
IP7
IP6
IP5
IP3
IP2
IP1
0
Operator ID
Figure 41: Comparison of % time spent in bucket fill vs. bucket empty – IP
Performing the analysis for the IP operators, it is noticeable that more IP operators spend a
higher percentage of time within the cycle time, filling the bucket rather than standing still. The
difference in percentage time spent in the case of those operators who are standing still more
than filling the bucket is much lesser than the IA operators which indicate that the IP operators
maximize the time in a cycle time by using it for productive work, as can be seen in figure 41.
70
45
40
35
Time (s)
30
25
20
% Bucket Fill Time
15
% Bucket Empty Time
10
5
IA18
IA17
IA16
IA15
IA14
IA13
IA12
IA11
IA10
IA8
IA6
IA5
IA4
IA3
IA2
IA1
0
Operator ID
Figure 42: Comparison of % time spent in bucket fill vs. bucket empty – IA
As can be seen from figure 42, more than 60% of the IA operators spend more time in standing
still than the time spent in filling the bucket. What is noticeable about these operators is that in
some of their cases, the time spent standing still is almost double the time spent in filling the
bucket. However in the case of the few operators who do spend more time filling the bucket, the
difference between the two percentage values is very less.
71
50
45
40
Time (s)
35
30
25
20
% Bucket Fill Time
15
% Bucket Empty Time
10
5
IR14
IR12
IR11
IR10
IR9
IR8
IR7
IR6
IR5
IR3
IR2
IR1
0
Operator ID
Figure 43: Comparison of % time spent in bucket fill vs. bucket empty – IR
The analysis of the cycle times for the IR operators reveals that the IR operators spend double
the percentage of time standing still as compared to the time spent in filling the bucket. As can
be observed from figure 43, the productive time for these operators is quite low compared to the
previous two classes of operators.
72
4.6.2 Cycle Times for Load & Carry (Gravel)
Boxplot of Cycle Times - Bandet
225
200
175
Data
150
125
100
75
IA1
IA2
IA3
IA4
IA5
IA6
IA8
IA10
IA11
IA12
IA13
IA14
IA15
IA16
IA17
IA18
IP1
IP2
IP3
IP5
IP6
IP7
IP8
IP9
IP10
IP11
IP12
IP13
IP14
IP15
IP16
IR1
IR2
IR3
IR5
IR6
IR7
IR8
IR9
IR10
IR11
IR12
IR14
50
Figure 44: Cycle Times for all operators – L&C (Gravel)
The above figure 44 shows the cycle times for the operators in the L&C application. This
application takes longer compared to the other 2 applications due to the transport in the L&C
application. As before, there is lesser variation between the IA and IP operators compared to the
IR’s. As highlighted in the plot, the best performing operator is IP2 again. This operator is more
consistent in performing the application with his cycle times showing lesser variation compared
to any of the other operators. The presence of a single outlier in his case indicates the last cycle
which he performed, which the time signal might not have correctly measured.
A similar analysis as in the case of SLC (Gravel) can be done here to identify the percentage
time spent in each phase of this application. It is quite obvious that a majority of the time in this
application is spent on the transport.
73
4.6.3 Cycle Times for Short Loading Cycle (Rock)
Boxplot of Cycle Times - Berg
160
140
Data
120
100
80
60
40
IA1
IA2
IA3
IA4
IA5
IA6
IA8
IA10
IA11
IA12
IA13
IA14
IA15
IA16
IA17
IA18
IP1
IP2
IP3
IP5
IP6
IP7
IP8
IP9
IP10
IP11
IP12
IP13
IP14
IP15
IP16
IR1
IR2
IR3
IR5
IR6
IR7
IR8
IR9
IR10
IR11
IR12
IR14
20
Figure 45: Cycle Times for all operators – SLC (Rock)
The SLC (Rock) application is an application which requires more skill compared to the SLC
(Gravel) application as it involves the loading of a heavier material. As before, operator IP2 is
the best performing operator with the least cycle time compared to the other operators. It can be
clearly observed in figure 45, that there is very little variation in cycle times within the IP
operators. This can be attributed to the higher skill possessed by these operators compared to the
other 2 classes of operators.
The analysis of the cycle times for all 3 applications shows that the IP operators perform the
required operations in the least time, followed by the IA operators and finally the IR’s. The fact
that the IP’s have spent more time operating the WLO compared to the other 2 classes is clearly
distinguishable here.
74
70
60
Time (s)
50
40
% Bucket Fill Time
30
% Bucket Empty Time
20
10
IP16
IP15
IP14
IP13
IP12
IP11
IP10
IP9
IP8
IP7
IP6
IP5
IP3
IP2
IP1
0
Operator ID
Figure 46: Comparison of % time spent in bucket fill vs. bucket empty – IP
Analyzing for the IP operators, as seen in figure 46, it can be observed that while for most
operators, the percentage of bucket fill time is more than the percentage of standing still time, in
a couple of cases, nearly 2.5 times, there are a few operators who spend an equal percentage of
time in both the phases. The fact that the IP operators are more experienced in handling this
application compared to the IA’s and the IR’s is reflected in the low variance among most of the
operators.
75
60
50
Time (s)
40
30
% Bucket Fill Time
20
% Bucket Empty Time
10
IA18
IA17
IA16
IA15
IA14
IA13
IA12
IA11
IA10
IA8
IA6
IA5
IA4
IA3
IA2
IA1
0
Operator ID
Figure 47: Comparison of % time spent in bucket fill vs. bucket empty – IA
The SLC (Rock) application is a harder task to perform compared to the SLC (Gravel)
application. The operators, no matter how skilled they are, will spend more time in filling the
bucket with rock as compared to the time needed to fill it with gravel. This is reflected in the
graph above where the percentage of time spent in bucket fill is now more than the percentage of
time spent with the bucket empty (standing still). In fact, in the case of some of the IA operators,
the percentage of bucket fill time is almost double the percentage of standing still time, as can be
seen in figure 47.
76
60
50
Time (s)
40
30
% Bucket Fill Time
20
% Bucket Empty Time
10
IR14
IR12
IR11
IR10
IR9
IR8
IR7
IR6
IR5
IR3
IR2
IR1
0
Operator ID
Figure 48: Comparison of % time spent in bucket fill vs. bucket empty – IR
Finally, in figure 48, for the IR’s, it can be observed that while most operators spend an equal
percentage of time in both the phases, their overall cycle time is longer than the previous classes
of operators. This can be attributed to their lack of experience on the WLO and hence the longer
cycle times.
77
Conclusions
We can thus conclude that the IP operators due to their WLO experience and time spent driving
the machine, take lesser time to complete a cycle while spending an equal percentage of time in
the bucket fill and bucket empty phase. In comparison to the IP’s, the IA operators take longer to
complete a cycle while displaying variation between the percentages of time spent in the two
phases. The IR’s take the most time to complete a cycle, quite obviously. Also, the level of
difficulty in the SLC (Rock) application causes it to be more time consuming than the SLC
(Gravel) application, with more time being spent in filling the bucket with rock compared to
gravel. Lastly, the percentage of time spent in transport is the highest in the L&C (Gravel)
application, compared to the other two applications.
78
4.7 Fuel Efficiency
The fuel efficiency and productivity of each operator in each of the 3 applications is as shown in
the graphs below. As can be observed, the productivity of the IP operators is much higher
compared to that for the IR operators, which can be expected due to the fact that the IP’s are the
most experienced among the classes of operators. It can be seen that the IP operators have a
productivity of over three to seven times than the IR operators, depending upon the application.
Also, the fuel efficiency is for the IP’s is over two to three times the IR’s. This establishes the
fact that the more experienced the operator, the better is his fuel efficiency and productivity
Fuel Efficiency (ton/l)
(Frank, B. et al., 2012) [2].
IP
IA
IR
Productivity (ton/hr)
Figure 49: Fuel efficiency vs. Productivity – SLC (Gravel)
In the SLC (Gravel) application, the IP’s and the IA’s are clustered closer to each other i.e., there
is lesser variation among these classes of operators, as can be seen in figure 49. The best
performing operator is this case is IP2 with the highest fuel efficiency value and the second best
productivity. The poor fuel efficiency and productivity values for the IR’s can be attributed to
their lack of experience.
79
Fuel Efficiency (ton/l)
IP
IA
IR
Productivity (ton/hr)
Figure 50: Fuel efficiency vs. Productivity – L&C (Gravel)
The L&C (Gravel) application requires more fuel for completion due to the transport involved in
it. Hence, the fill factor is important here, the operators need to ensure that each bucket was filled
as best as they could, to have a high fuel efficiency and productivity. In figure 50 too, it can be
observed that the IP’s and the IA’s are clustered closer together, with operator IP1 being the best
performing operator here.
80
Fuel Efficiency (ton/l)
IP
IA
IR
Productivity (ton/hr)
Figure 51: Fuel efficiency vs. Productivity – SLC (Rock)
Lastly, in the SLC (Rock) application, the degree of difficulty in filling the bucket due to the
nature of the material being filled causes a larger deviation among the operators. In this
application specifically, the experience of the operator played a key role in how well he
performed. Operator IP2 is yet again the best performing operator by a clear margin over the
other operators. As can be noticed by the figure 51, some IR operators performed better than few
of the IA operators!
81
The operators were posed the question as to how good of a fuel efficiency did they get in the
rock and gravel handling applications. Their answers (rated 1 – 4) were then compared to their
performance to identify which class of operators performed the best and if there were any
clusters forming among them.
In the SLC (Rock) application, 27 of the 43 operators interviewed were not satisfied with the fuel
efficiency they achieved during the operation of the WLO. One of the reasons for this could be
the level of difficulty involved in this application. Figure 52 shows the values of fuel efficiency
for each operator in this application. While being interviewed, these operators gave lack of time
spent on the machine and need to improve their handling of the WLO as reasons of
dissatisfaction rather than the machine being incapable of delivering a good productivity with
IA1
IA2
IA3
IA4
IA5
IA6
IA8
IA10
IA11
IA12
IA13
IA14
IA15
IA16
IA17
IA18
IP1
IP2
IP3
IP5
IP6
IP7
IP8
IP9
IP10
IP11
IP12
IP13
IP14
IP15
IP16
IR1
IR2
IR3
IR5
IR6
IR7
IR8
IR9
IR10
IR11
IR12
IR14
Fuel Efficiency (ton/l)
good fuel efficiency.
Operator ID
Figure 52: Fuel Efficiency for all operators – SLC (Rock)
82
Fuel Efficiency (ton/l)
IA1
IA2
IA3
IA4
IA5
IA6
IA8
IA10
IA11
IA12
IA13
IA14
IA15
IA16
IA17
IA18
IP1
IP2
IP3
IP5
IP6
IP7
IP8
IP9
IP10
IP11
IP12
IP13
IP14
IP15
IP16
IR1
IR2
IR3
IR5
IR6
IR7
IR8
IR9
IR10
IR11
IR12
IR14
Operator ID
Figure 53: Fuel Efficiency for all operators – SLC (Gravel)
In the SLC (Gravel) application, only 4 of the operators were dissatisfied with the fuel efficiency
and all of them belong to the IR class. Again, these operators felt that it was more to do with
personal skill and lack of time spent on the WLO rather than the machine being poor as the
reason for their low fuel efficiency ratings, as can be seen by the red lines in figure 53.
Conclusions
A quite obvious conclusion drawn here is that the IP operators, due to their experience perform
more fuel efficiently and have a higher productivity compared to the other classes of operators.
This can be seen in the above presented graphs where the IP’s are invariably grouped together at
the farther end of the scatter graph which shows higher values of fuel efficiency and
productivity.
Also, most of the IA and IR operators felt that with more training and time spent on the WLO,
they could improve both the parameters considerably, thus lessening the gap between themselves
and the IP’s.
83
4.8 Rimpull
Different definitions of rim pull exist in the construction industry today. Some of them are stated
below:
“The friction forces, expressed in pounds, exerted by rubber tires on driving wheels and the
contact surface”
- Dictionary of Construction.com
“The available rim pull can be defined as the force that the drive wheel delivers to the ground in
order to propel the vehicle”
- Mine Planning and Equipment Selection, 1996
Quite simply, the rim pull is the usable torque at the point of contact between the tires and the
ground for a wheel machine. The operators were asked if they felt the machine had enough rim
pull during the measurements conducted. All the answers received for this question were positive
with 24 operators responding with a rating of 3 and 19 responding with a rating of 4. None of the
operators interviewed felt that the machine did not possess enough rim pull. In the light of this, a
general analysis of rim pull calculations has been done to show how the rim pull varies while
using the 1st and 2nd gear and the various parameters involved in it.
The parameters required to calculate the rim pull are:
-
Output torque from the torque converter
-
Mechanical efficiency of the pinion and final axle drive
-
Mechanical efficiency of the clutch + gear
-
Rolling radius
-
Gear Ratio
-
Ratio of engine speed to pump speed called slip
-
Ratio of engine torque to pump torque
The values of the pump torque at a 1000 RPM are obtained from the supplier of the converter for
different values of slip.
84
This is then used to calculate the pump torque at different RPM’s by using the formula:
!"#
= $%%%!"# ∗ (
'()*+, .
) 1000+,
This obtained torque is the “output torque” from the torque converter. It is then used to calculate
the rimpull at different values of slip at different RPM’s by using the below formula:
+/** = ((
0121340 ∗ )5063053&898 ∗
,(ℎ)()*;;((<8313&=3>?>@?
4
∗
,(ℎ)()*;;((<2?02ABC> ∗
D)
+))/+**
)E/1000) ∗ D)
F)(
The units of rim pull are kN (kilo Newton).
Since the operators are satisfied with the rim pull that the machine offers, there is no comparison
point between their answers and the measurements. A general graph of how the rim pull varies
for each value of slip in gear 1 is presented below, in figure 54.
0,1
0,2
Rimpull (kN)
0,3
0,4
0,5
0,6
0,7
0,8
0,9
2100
1900
1850
1800
1700
1600
1500
1400
1300
1200
1150
1100
1050
1000
900
800
0,95
0,973
Engine Speed (RPM)
Figure 54: Variation in rim pull for different slip values in gear 1
85
As can be seen from the above graph, the rim pull value progressively increases as the number of
RPM’s in the engine goes up. The graph shows the values of rim pull starting from the value of
0.1 slip at the top, down to the value of 0.973 slip at the bottom of the graph. Conversely, the rim
pull reduces as the amount of slip between the engine and the pump increases.
0,1
0,2
Rimpull (kN)
0,3
0,4
0,5
0,6
0,7
0,8
0,9
2100
1900
1850
1800
1700
1600
1500
1400
1300
1200
1150
1100
1050
1000
900
800
0,95
0,973
Engine Speed (RPM)
Figure 55: Variation in rim pull for different slip values in gear 2
The same calculations can be done for the machine being driven in gear 2 as well. As before, it
can be observed in figure 55 that the rim pull increases as the engine speed increases. There is a
positive correlation between the rim pull and the engine speed. And again, as the slip value
increases, the rim pull value decreases i.e. there is a negative correlation between the two.
To see how the operators perform and how much rim pull they utilize, one phase in one
application was chosen as an example to highlight if it is possible to cluster the operators
according to their usage. The phase and application considered here is the bucket fill phase in the
SLC (Rock) application. The method can similarly be extended for other phases in the other
applications too.
86
Clustering of rimpull values
Similarity
57,77
71,84
100,00
1
22
24
4
27
37
40
20
41
9
12
17
30
19
33
13
7
15
8
34
36
38
42
43
6
23
26
11
21
35
31
5
25
2
10
28
16
32
18
29
3
39
14
85,92
Operator Number
Figure 56: Clustering of rim pull used by operators in bucket fill phase for SLC (Rock) in gear 1
As can be seen from the above dendogram (figure 56), most of the operators have a very high
similarity level in terms of rim pull usage. There still are a few operators who do not utilize as
much rim pull as the machine offers, but they still are satisfied with the WLO. Attached in the
appendix, are the steps that the program, Minitab, performs the clustering in. Also, as can be
seen in figure 57, the grouping of the operators is quite close with just 3 operators separated from
the main group, wherein lie most of the operators.
87
Maximum Rimpull (kN)
IP
IA
IR
Minimum Rimpull (kN)
Figure 57: Operator grouping based on range of rim pull used in the bucket fill phase in SLC
(Rock) in gear 1
As stated before, this methodology can be extended to other phases within an application as well
as to other applications to see how different operators utilize different rim pull at differing slip
values.
Conclusions
The conclusions drawn from the analysis of the rim pull are that most operators utilize as much
rim pull that the machine has available and hence respond that they are satisfied with the rim
pull. What can also be seen is that the rim pull available at gear 1 is much higher than that
available at gear 2. The scatter plot shown above does not really give a clear idea as to what
classes of operators utilize the most available rim pull as all three classes of operators are
grouped very close to each other.
88
Chapter III
89
5 Clustering
This chapter deals exclusively with the clustering done in this thesis to see if any patters or
groupings emerge from the analysis of the answers given to the questionnaire by the operators
and their actual performance measures. This is done to be able to verify if operators from a
particular class utilize the WLO in the same manner or if they prefer certain
parameters/characteristics of the machine over others. This will help Volvo CE to tailor their
products to suit particular customer segments, based on the feedback received.
Originally, the clustering available in Minitab was used to see if any results emerge for the
clustering. But, since Minitab offers only dendograms (as used in an above section), a different
tool was used for the clustering. A Microsoft Excel add-in, NeuroXL Clusterizer was used to
cluster the results which have been presented below. For all the clustering done, three clusters or
groups were chosen to see if the tool clusters the operators based on their classes i.e. IP, IA and
IR or does any other form of grouping emerge. The reason more clusters were not used in the
tool is because in some cases, only one operator was assigned in a cluster, which did not reveal
any usable result.
The most commonly used clustering methods are Partitional and Hierarchical techniques (Han
and Kamber, 2001). For this report, the Hierarchical clustering technique is used. Within this
method, there are two approaches that can be used, agglomerative (bottom-up) and divisive (topdown). NeuroXL utilizes the agglomerative method wherein, the clusterizer starts with each
object being a separate cluster by itself and then successively merges or clusters groups
according to their similarity. The clustering finishes when all similar objects are in the same
cluster and dissimilar objects are in different clusters (Andritsos, 2002).
90
5.1 Clustering of accuracy (difference between answered and actual values from Chapter I)
To identify if any clusters emerge from how accurate the operators are in estimating their
performance on the machine, the difference between their answers and the actual value was
taken and used in the clustering tool. Figure 58 indicates how the clusters are formed. As can be
seen, there is lesser difference between the three clusters for the 1st and the 5th parameter. As can
be seen, there are no clusters forming for parameter 3 and 4, this is because the values are so
close to 0 that the clustering tool does not highlight them on the graph.
Clusters profiles
Cluster 1
Cluster 2
Cluster 3
150.00%
100.00%
50.00%
0.00%
Rock (RPM)
Gravel (RPM)
SLC (km/hr)
L&C (km/hr)
-50.00%
SLC Distance
(m)
-100.00%
-150.00%
-200.00%
-250.00%
Figure 58: Cluster Profile for accuracy for all operators
The clustering as attached in the appendix shows that the clusters are heterogeneous i.e. each
cluster is made up of operators from all the 3 classes, there is no one cluster made up exclusively
of operators from a certain class. Also, cluster 3 is made up of more IP operators than from the
other two classes. This cluster is made up of operators who are quite accurate in their estimations
of their performance on the machine. The difference between their answers and the actual usage
is a lot lesser in 3 of the 5 parameters considered here, as compared to the operators from the
other two classes.
The below graph (figure 59) for the cluster weights, show that more operators belong to cluster
2. This cluster is made up of more IA’s and IR’s, with fewer IP’s. This is a cluster made up of
those operators who are not the most accurate in estimating the engine speed at which they
91
performed the bucket fill phase, but are more accurate in estimating the vehicle speed at which
they drove the machine during the transportation phase of the applications.
Clusters weights
45.00%
41.86%
40.00%
34.88%
35.00%
30.00%
25.00%
23.26%
20.00%
15.00%
10.00%
5.00%
0.00%
1
2
3
Figure 59: Cluster weights for each cluster from chapter I
The conclusions that can be drawn from this are that most of the IP operators (belonging to
cluster 3) have a good estimate of how they operated the machine, whereas most of the IA’s and
IR’s though have a good estimate, are nowhere as close to the IP’s.
Difference
Difference between
Difference between
between
answered and actual
answered and actual
answered and
actual
Rock (RPM) Gravel (RPM) SLC (km/hr) L&C (km/hr) SLC Distance (m)
163.1
66.5
-2
-5
14.705
222.0555556 72.72222222 -0.777777778
-1.5
8.916666667
115.2666667 -38.66666667 -3.133333333 -2.733333333
3.756
Cluster Average
Cluster 1 average
Cluster 2 average
Cluster 3 average
Table 5: Average values for the clusters for parameters from chapter I
92
5.2 Clustering of preferences from chapter II
In this section of the report, a clustering of the operators’ preferences was done to be able to
identify any patterns emerging from it. Here, the preferences of the operators, by their ratings,
are clustered, the parameters from chapter II and the intangible parameters from chapter III. As
can be seen from the below cluster profile graph (figure 60), the least difference among the 3
clusters of operators is for parameter 11, which is the rating for rim pull, which incidentally is
the one question where all the interviewed operators felt that the machine had good rim pull.
Clusters profiles
Cluster 1
Cluster 2
Cluster 3
15.00%
10.00%
5.00%
0.00%
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
-5.00%
-10.00%
-15.00%
-20.00%
Figure 60: Cluster Profile for preferences for all operators
The results of the clustering are included in the appendix and they show that cluster 1 is made up
of mostly the IP operators who overall seem quite satisfied with the machine performance with
respect to most of the parameters considered. The one area of concern as highlighted by the
analysis in previous sections is the gradeability i.e. the power required by the WLO to climb the
hill during the transportation phase of the L&C application, but this is one parameter which has a
negative response in cluster 1 and cluster 2, whereas cluster 3, made up of more IA’s and IR’s
seem satisfied with it.
93
Clusters weights
45.00%
40.00%
39.53%
37.21%
35.00%
30.00%
23.26%
25.00%
20.00%
15.00%
10.00%
5.00%
0.00%
1
2
3
Figure 61: Cluster weights for each cluster from chapter II
The above cluster weights graph (figure 61) shows that more operators belong to cluster 1 and 3.
As can be observed from this clustering, again, the clusters are not formed according to the
classes of the operators, though cluster 1 is dominated by the IP operators and cluster 3 is
heterogeneously made up of both IA’s and IR’s.
The conclusions drawn from this section are that cluster 1 is made up of operators who are better
able to recognize the lack of engine power (gradeability) which is not so surprising seeing that
this cluster has more IP operators than the other two clusters, while they also rate the lifting force
available as quite low. On the other hand, cluster 3, made up of more IA’s and IR’s seem
satisfied with most of the parameters considered in chapter II.
94
5.3 Clustering of parameters from chapter II using weighted scores
This section deals with the clustering of the parameters from chapter II after assigning specific
weights to them. The weight’s percentage was decided after receiving feedback from the
operators during the interviews as to how they valued the parameters from chapter II. The author
of the thesis has also made an assumption as to the importance of these parameters while
assigning the weights to them. The assigned weightage is as below:
Fuel Efficiency – 30%
Rim pull – 20%
Lifting force – 15%
Breakout force – 10%
Power (Gradeability) – 15%
Overall power – 10%
Multiplying the ratings given by the operators by their assigned weights and then adding all the
multiplied weights to get a total weighted scored, the clustering was then done on these total
weighted scores.
Clusters profiles
Cluster 1
Cluster 2
Cluster 3
20.00%
15.00%
10.00%
5.00%
0.00%
-5.00%
1
2
3
4
5
6
7
8
9
10
-10.00%
-15.00%
-20.00%
-25.00%
-30.00%
Figure 62: Cluster profiles for weighted scores for parameters from chapter II
95
As can be seen from figure 62, the variation among the clusters is the least for parameters 3 (Rim
pull) and 9 (overall power). The clustering is as attached in the appendix and it can be clearly
seen that the clusters are very heterogeneous; no one cluster is made up of purely one class of
operators. There is very less variation between the average values of the total weighted scores for
the 3 clusters, as seen in table 6.
Fuel Efficiency
Rating (30%)
Power
Total
Rim pull Lifting Force (15%) Breakout Force (10%)
Overall Power
(Gradeability)
Weighted
Rating (20%)
(10%)
(15%)
Score
Gravel
Rock Gravel
Rock
Gravel
100
68
45
51 32.666667
34
40
30.66666667 475.3333333
87.5 66.66666667 45
48.75 29.166667 35.833333
37.5
32.5
430.4166667
90
71.25
53.4375 55.3125 32.5
38.125
45
32.5
485.625
Cluster Average
Rock
74
47.5
67.5
Cluster 1 average
Cluster 2 average
Cluster 3 average
Table 6: Average values of weighted scored for parameters from chapter II
In this clustering, the 3 clusters are spread out such that, most of the operators fall into cluster 3,
followed by cluster 1. What is noticeable is that cluster 1 is made up of more IP’s and cluster 3 is
made up of more IA’s than other classes.
This clustering does not reveal much except the fact that the clusters are heterogeneous and that
the differences between the 3 clusters are quite less when using the weighted score method.
96
5.4 Clustering of actual values of parameters from chapter II
In this section of the report, the actual values of the parameters from chapter II are clustered to
see if any patterns or groupings emerge. For some parameters, absolute values have been taken
because the algorithm used in Matlab to record these values only records a single value (such as
for fuel efficiency and productivity), whereas for other parameters, the maximum recorded value
has been chosen (rim pull and lifting force). In the other remaining cases, either a median value
(output torque for gradeability and overall power) or an average value (for bucket loads) has
been chosen, depending upon how the data for that parameter is distributed (normal or notnormal).
Clusters profiles
Cluster 1
Cluster 2
Cluster 3
30.00%
20.00%
10.00%
0.00%
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
-10.00%
-20.00%
-30.00%
-40.00%
-50.00%
Figure 63: Cluster profiles for actual values of parameters from chapter II
What is noticeable from figure 63 is that cluster 1 has an above average usage of the parameters
such as rim pull, lifting force, output torque with a higher fuel efficiency, productivity and
average bucket load filled. It is not surprising to see that this cluster is made up of all the IP
operators and a few other IA operators. There are no IR’s in this high performing cluster. There
are a few IR’s in cluster 2 whereas cluster 3 is made up of only IR’s.
It is also noticeable to see that the cluster 3 is made up of those operators who have the least
average values for all the parameters.
97
Clusters weights
70.00%
60.00%
58.14%
50.00%
40.00%
30.23%
30.00%
20.00%
11.63%
10.00%
0.00%
1
2
3
Figure 64: Cluster weights for each cluster of actual values from chapter II
As can be seen from figure 64, more than half the operators belong to cluster 1, which is the
cluster made up of high performance operators.
There is very little difference between the 3 clusters in parameters 7 (rim pull) and 14 (average
bucket load in L&C – Gravel). This can be attributed to the fact that most operators utilized the
same amount of rim pull while operating the machine and hence it influenced their answers of
satisfaction while giving a rating. The close similarity among the average bucket loads for the
L&C application can be attributed to the fact that the operators felt the need to fill up the bucket
as much as possible in every cycle to be more fuel efficient, since this application involved a
longer transportation phase in it. Also, since the material was constantly being recycled, into a
hopper which then delivered the material back to the gravel pile, the amount of material
remained the same, ensuring that the operators almost always filled up the bucket as full as they
could.
98
5.5 Clustering of combination of preferences and actual values from chapter II
The final clustering in this report is done by combining the preferences and the actual usage
values of the operators for each parameter from chapter II. This is done to be able to see, overall,
if any patterns or grouping emerge from the answers to the questionnaire and the actual usage of
the WLO by the operators. This clustering is the largest clustering in this report, due to the
volume of data involved.
Clusters profiles
Cluster 1
Cluster 2
Cluster 3
30.00%
20.00%
10.00%
0.00%
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31
-10.00%
-20.00%
-30.00%
-40.00%
-50.00%
Figure 65: Clustering of preferences and actual usage values from chapter II
As can be seen from figure 65, cluster 1 and cluster 2 are more similar in terms of their
preferences and usage than cluster 3. What is again noticeable is that cluster 1 is made up of a
majority of the IP’s and some IA’s. This is a cluster made up of the high performing operators
who are quite happy with performance of the machine too. On the other hand, cluster 3, made up
of only the IR’s are the poorer performing operators with lower expectations about the machine
performance.
99
Clusters weights
50.00%
46.51%
45.00%
40.00%
35.00%
30.23%
30.00%
23.26%
25.00%
20.00%
15.00%
10.00%
5.00%
0.00%
1
2
3
Figure 66: Cluster weights of each cluster for preferences and actual usage values from chapter II
It is also noticeable to see in figure 66 that close to half of the operators interviewed, belong to
cluster 1. What can be seen from the clustering here is that, the average values for most of the
parameters, preferences and actual usage, are higher for cluster 1 and 2 compared to cluster 3.
The conclusions drawn from the clustering are that most IP’s are satisfied with the machine
performance in terms of their preferences and also utilize the WLO to a high capacity while
performing the different applications, whereas the IR’s on the other hand, have lower preferences
as well as do not utilize the machine that well. Again, in this clustering, it can be seen that the
clusters formed are heterogeneous for cluster 1 & 2 and homogeneous with only rookies in
cluster 3.
100
6 Conclusions
As highlighted before, customer feedback is very important in new product development as it
allows the company to incorporate the customer’s ideas and suggestions into product design and
development. This involvement of the customer, called co-creation by many research authors,
gives the customer a sense of being involved in the company’s product development decisions,
thereby allowing the company to tailor a product or service to better suit the customer.
Design questionnaire from
literature review and previous
questionnaires used by Volvo CE
Conduct interviews and collect
feedback after choosing
interviewees from all levels of
organization
Identify what measurements and
signals are useful and extract
data from them
Analyze feedback from
interviews and from product
usage measurements
Cluster customers according to
preferences and usage patterns
Figure 67: Flowchart of methodology developed during the thesis work
101
Satisfying the employees who are the internal customers, before dealing with the external
customers, allows a company to focus on internal marketing and thereby improve their external
marketing of the product or service. Figure 67 presents the flowchart of the methodology
developed during this thesis work. The process as described by the flowchart can be utilized
further by Volvo CE while incorporating the ideas developed by the thesis author in future
customer feedback and data analysis projects.
One of the main goals of this thesis was to develop a methodology which combined the
preferences of the customer (internal operators in this case) with the actual usage of the WLO.
The literature survey done in the beginning of the thesis paved the way to be able to choose
interviews as the methodology to collect customer feedback. In spite of being a reactive method
of feedback collection, this method was chosen to be able to get both, the spoken and the
unspoken needs of the customers from their feedback to the interviews. This was done to be able
to have a loop of feedback that would ultimately impact any current and future design
requirements. The operators chosen were the same who participated in the previous measurement
in September 2011.
By acquiring feedback from the operators in the form of one-on-one interviews, their answers to
a questionnaire, which included the actual usage of the WLO as well as their preferences, was
captured and analyzed to identify any gaps that occur in the way that they handle the WLO and
whether this ultimately caused them to answer the questionnaire, the way they did. This also
reveals an obvious trend that the IP’s are a lot more accurate at estimating their performance on
the WLO as compared to the other classes of operators.
A deeper analysis of the input signals reveals what measurements are useful to be able to analyze
the data collected from the interviews and the product usage measurements. This was a more
challenging task due to the sheer volume of the data captured by the signals. There is now a
clearer picture for Volvo CE as to what signals and particularly what data phases are important to
be able to identify the useful data from the operator feedback and actual usage. The analysis of
the data gives an idea that the IP operators are better at using the machine and have more
reasonable preferences as compared to the IR’s who are poorer at utilizing the WLO and have
higher expectations from the machine. In the author’s opinion, this can be attributed to the
experience and more time spent driving a WLO for the IP’s compared to the IA’s and IR’s.
102
Finally, the clustering of the data as covered in the final chapter reveals a more heterogeneous
grouping in most cases, where clusters 1 are made up of those operators who are better at using
the WLO and have reasonable preferences from the machine, whereas clusters 3 are invariably
made up of those operators who are quite inexperienced at handling the WLO and hence have
higher preferences from the machine, to make up for their lack of skill or time spent on a WLO.
To summarize, choosing interviews as the method to collect customer feedback from the internal
customers allowed for a correlation to be drawn between the answers to the questionnaire in the
interviews and the actual usage of the machine by the customers. An analysis of the data
collected is then used to be able to group the customers according to their preferences and how
the utilize the WLO. This will allow Volvo CE to tailor products to suit individual market
segments.
103
7 Future Works
As highlighted before, if Volvo CE deems the methodology developed and utilized in this thesis
as successful, they can then roll out the questionnaire developed to external customers, to be able
to capture and analyze the data in the same manner as done in this report. This will allow Volvo
CE to cover a wide base of customers whose feedback can prove to be invaluable while making
decisions regarding design changes and for new product development.
Also, this methodology can then be utilized for other variants of the WLO ranging from the
small L15 to the large L350 types. The success of this methodology will also allow Volvo CE to
use it for other construction equipment they manufacture such as articulated haulers and
excavators, to be able to better capture and analyze customer feedback.
104
8 References
Alam, I. (2002), “An exploratory investigation of user involvement in new service
development”, Journal of the Academy of Marketing Science, Vol.30 No.3, pp. 250-261.
Bergman, B. and Klefsjö, B. (2003), Quality – From Customer Needs to Customer Satisfaction,
Studentlitteratur, Lund.
Bouchereau, V. and Rowlands, H. (2000), “Methods and techniques to help quality function
deployment (QFD)”, Benchmarking: An International Journal, Vol.7 Issue 1, pp. 8–20.
Burnard, P. (1991), “A method of analyzing interview transcripts in qualitative research”, Nurse
Education Today, Vol.11 Issue 6, pp. 461-466.
Eisenhardt, K.M. (1989), “Building theory from case study research”, The Academy of
Management Review, Vol.14 No.4, pp. 532-550.
Elg, M. (2007), “The process of constructing performance measurement”, The TQM Magazine,
Vol.19 No.3, pp. 217-228.
Frank, B., Skogh, L. and Alaküla, M. (2012), “On wheel loader fuel efficiency difference due to
operator behavior distribution”, 2nd Commercial Vehicle Technology Symposium (CVT2012),
Kaiserslautern, Germany, March 13-15, 2012.
Gill, P., Stewart, K., Treasure, E. and Chadwick, B. (2008), “Methods of data collection in
qualitative research: interviews and focus groups”, British Dental Journal, Vol.204 No.6, pp.
291-295.
Gounaris, S. (2008), “The notion of internal market orientation and employee job satisfaction:
some preliminary evidence”, Journal of Services Marketing, Vol.22 No.1, pp. 68-90.
Grönroos, C. and Ravald, A. (2011), “Service as business logic: implications for value creation
and marketing”, Journal of Service Management, Vol.22 No.1, pp. 5-22.
Grönroos, C. (2011), “Value co-creation in service logic: A critical analysis”, Marketing Theory,
pp. 279-301.
105
Gruner, K.E. and Homburg, C. (2000), “Does Customer Interaction Enhance New Product
Success?”, Journal of Business Research, Vol.49 No.1, pp. 1-14.
Gustafsson, A., Ekdahl, F. and Edvardsson, B. (1999), “Customer focused product development
in practice – a case study at Scandinavian Airlines System (SAS)”, International Journal of
Service Industry Management, Vol.10 No.4, pp. 344-358.
Gustafsson, A., Ekdahl, F., Falk, K. and Johnson, M. (2000),”Linking Customer Satisfaction to
Product Design: A Key to Success for Volvo”, Quality Management Journal, Vol.7 No.1, pp. 2738.
Hirschman, E.C. (1986), “Humanistic Inquiry in Marketing Research: Philosophy, Method and
Criteria”, Journal of Marketing Research, Vol.23, pp. 237-249.
Jaworski, B.J., Kohli, A.K. and Sahay,A. (2000), “Market-driven versus driving markets”,
Journal of the Academy of Marketing Science, Vol.28 No.1, pp.45-54.
Johnson, M.D. (1998), Customer Orientation and Market Action, Prentice-Hall, Upper Saddle
River, NJ.
Kollberg, B., Elg, M. and Lindmark, J. (2005), “Design and implementation of a performance
measurement system in Swedish health care services: a multiple case study of six development
teams”, Quality Management in Health Care, Vol.14 No.2, pp. 95-111.
Kristensson, P., Gustafsson, A. and Archer, T. (2004), “Harnessing the creative potential among
users”, Journal of Product Innovation Management, Vol.21 No.1, pp. 4-14.
Kristensson, P., Gustafsson, A. and Witell, L. (2011), “ Collaboration with Customers –
Understand the Effect of Customer-Company Interaction in new Product Development”,
Proceedings of the 2011 44th Hawaii International Conference on System Sciences, pp. 1-9.
Lucas, C. and Kline, T. (2008) "Understanding the influence of organizational culture and group
dynamics on organizational change and learning", The Learning Organization, Vol. 15 Iss.3, pp.
277 – 287.
106
Narver, J.C., Slater, S.F. and MacLachlan, D.L. (2004), “Responsive and proactive market
orientation and new product success”, Journal of Product Innovation Management, Vol. 21
No. 5, pp. 334-47.
Neely, A., Gregory, M. and Platts, K. (1995), “Performance measurement system design – a
literature review and research agenda”, International Journal of Operations and Production
Management, Vol.15 No.4, pp. 80-116.
Payne, A.F., Storbacka, K. and Frow, P. (2008), “Managing the co-creation of value”, Journal of
the Academy of Marketing Science, Vol. 36 No. 1, pp. 83-96.
Roberts-Lombard, M. (2010), “Employees as customers – An internal marketing study of the
Avis car rental group in South Africa”, African Journal of Business Management, Vol. 4(4), pp.
362-372.
Trott, P. (2001), “The role of market research in the development of discontinuous new
products”, European Journal of Innovation Management, Vol. 3, pp. 117-125.
Vargo, S.L. (2008), “Customer integration and value creation – paradigmatic traps and
perspectives”, Journal of Service Research, Vol. 11 No. 2, pp. 211-215.
Verma, R., Anderson, C., Dixon, M., Enz, C., Thompson, G. and Victorino, L. (2008), “Key
elements in service innovation: insights for the hospitality industry”, Cornell Center for
Hospitality Research Managerial Reports, Vol. 8 No. 18.
Voima, P., Heinonen, K. and Strandvik, T. (2011), “Value in experience – Proposing a customer
dominant marketing vocabulary”, EMAC 40th Conference, 24-27th May, 2011, Ljubljana.
Witell, L., Kristensson, P., Gustafsson, A. and Löfgren, M. (2011), “Idea generation: customer
co-creation versus traditional market research techniques”, Journal of Service Management,
Vol.22 No.2, pp. 140-159.
107
Books
Adèr, H.J. and Mellenbergh, G.J. (contributions by Hand, D.J.) (2008), “Advising on Research
Methods: A consultant's companion”, Huizen, the Netherlands: Johannes van Kessel Publishing.
De Vaus, D. (2002), “Surveys in Social Research”, 5th ed. Oxon:Taylor and Francis Ltd.
Denscombe, M. (2007), “The Good Research Guide”, 3rd ed. Maidenhead: McGraw-Hill.
Flodin, S., Nelson, T. and Gustafsson, A. (1997), "Improved Customer Satisfaction is a Volvo
Priority", in Johnson, M.D., Herrmann, A., Huber, F. and Gustafsson, A (eds) Customer
Retention in the Automotive Industry Gabler, Wiesbaden.
Han, J. and Kamber, M. (2001), “Data Mining: Concepts and Techniques”, Morgan Kaufmann
Publishers.
Jain, A.K. and Dubes, R.C. (1988), “Algorithms for clustering data”, Prentice-Hall.
Jarke, M., Lenzerini, M., Vassiliou, Y. and Vassiliadis,P. (1999), “Fundamentals of Data
Warehouses”, Springer.
Smith, D. B. (2012), “The People Make the Place: Dynamic Linkages Between Individuals and
Organizations”, Taylor & Francis.
Yin, R.K. (2008), “Case Study Research: Design and Methods”, 4th ed. Thousand Oaks: Sage
Publications, Inc.
Internet Links for references
[1]
http://www.volvoce.com/CONSTRUCTIONEQUIPMENT/CORPORATE/ENGB/ABOUTUS/Pages/about_us.aspx (2013-01-08)
[2]
http://www.iea.lth.se/publications/Papers/Frank_2012.pdf (2013-01-08)
[3]
http://www.purplemath.com/modules/boxwhisk.htm (2013-01-08)
Others
Andritsos, P. (2002), “Data Clustering Techniques”, University of Toronto, Department of
Computer Science.
108
Appendices
Questionnaire formulated and used during the interviews
1. At what RPM should you operate the WLO most fuel efficiently during bucket fill:- Rock Handling
Productivity
Efficiency
- Gravel handling
Productivity
Efficiency
2. At what maximum speed should you be driving the machine during transport for:- Short Loading Cycle
- Load and Carry
3. How good of a fuel efficiency do you get in:- Rock Handling
Full
Overfill
- Gravel Handling
Full
Overfill
Fill Factor
Cycle Time
4. Do you feel that the machine has enough rimpull?
5. Do you think that the lifting force is enough for:- Rock Handling
- Gravel handling
6. Do you think that the breakout force is enough for:- Rock handling
- Gravel
7. Do you think that the time is enough for:- Lifting
- Emptying
8. In short cycle loading, how long of a distance, in terms of wheel revolutions/meters/machine
length, did you travel per cycle?
9. Do you feel that the machine has enough power while climbing a hill during load and carry?
(Gradeability)
10. How do you feel about the machine stability with load:- Travelling straight forward
- Travelling while turned
11. Do you think that the engine is powerful enough?
109
12. Do you think that :- Reach
Lifting height of the vehicle is enough?
Intangibles
1. What are the major advantages of the L220?
Conversely, what are the major disadvantages/problems of the L220?
2. Why would you choose a Volvo machine?
Conversely, why would you choose a competitor’s machine?
3. Describe your CAB experience in terms of:- Sound
- Visibility
- Comfort level
- Boom Suspension system
- Gear shift jerk
- Feeling of the product
Vibrations
4. What features above the current ones, would you want in the product?
- Functionally
-
Aesthetically
5. How much seat time do you have on the product?
110
Clustering method followed in Minitab
Step
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
Number of
clusters
42
41
40
39
38
37
36
35
34
33
32
31
30
29
28
27
26
25
24
23
22
21
20
19
18
17
16
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
Similarity
level
99,4227
99,2649
99,2332
98,8755
98,7956
98,7406
98,7389
98,7313
98,6400
98,5925
98,5242
98,4554
98,3645
98,3452
98,0194
97,9442
97,8328
97,8020
97,7494
97,4386
97,4082
97,4033
97,3884
97,3352
97,2298
97,1407
97,0318
96,9890
96,9710
96,8594
96,7038
96,4098
96,2464
95,7940
95,6694
95,5099
95,2792
95,2566
95,2262
92,9733
91,5011
57,7668
Distance
level
1,1006
1,4015
1,4619
2,1439
2,2962
2,4010
2,4043
2,4188
2,5928
2,6834
2,8136
2,9448
3,1182
3,1548
3,7760
3,9194
4,1318
4,1906
4,2907
4,8834
4,9412
4,9506
4,9790
5,0805
5,2814
5,4512
5,6588
5,7404
5,7748
5,9875
6,2842
6,8448
7,1563
8,0188
8,2563
8,5605
9,0002
9,0433
9,1012
13,3963
16,2032
80,5175
Clusters
joined
17
30
27
37
4
27
21
35
36
38
9
12
4
40
7
15
34
36
42
43
21
31
9
17
16
32
3
39
9
19
7
8
4
20
10
28
6
23
4
41
3
14
10
16
11
21
9
33
4
9
34
42
10
18
10
29
4
13
22
24
1
22
4
7
1
4
1
34
1
6
1
26
1
11
5
25
1
5
2
10
1
2
1
3
New
cluster
17
27
4
21
36
9
4
7
34
42
21
9
16
3
9
7
4
10
6
4
3
10
11
9
4
34
10
10
4
22
1
4
1
1
1
1
1
5
1
2
1
1
Number
of obs.
in new
cluster
2
2
3
2
2
2
4
2
3
2
3
4
2
2
5
3
5
2
2
6
3
4
4
6
12
5
5
6
13
2
3
16
19
24
26
27
31
2
33
7
40
43
111
Using median value of the data set
The data mining and analysis in the report makes use of the median values for most of the data,
rather than the average. The box plots created further in the report have been done to be able to
get the median value in the data. For data which is not symmetrically distributed, the median is a
better reflection of the data compared to the mean as it gives a better idea of any general
tendency in the data.
The median value divides the data set (once ordered) into two halves. 50% of the data is above
and below it. If the number of data points is odd, the median value is an actual value that belongs
to the data set, if the number of data points is even, the average of the middle two data points is
the median value. Once this is done, again, the median values of the two halves (called as submedians, in some quarters) are taken, to give the “quartiles”, quarters into which the entire data
set is divided.
The main aim of using a box plot is to show how the distribution of the values in the data set is.
The IQR (Inter Quartile Range) can be used to measure this. Quite simply, the IQR is the
difference between the two sub-medians. It also shows the points which are considered to be too
far away from the central or median value. These points which lie far away from the central
value are called outliers, because they lie outside the range of data we work with. Assuming that
the sub-medians are called Q1 (bottom half) and Q3 (top half), the below formulae can be used to
calculate which values are outliers: Q1 – 1.5*IQR and Q3 + 1.5*IQR (www.purplemath.com).
112
6.1 Clustering of accuracy (difference between answered and actual values from Chapter I)
Operator ID
IP1
IP13
IP15
IA5
IA6
IA14
IR2
IR7
IR9
IR10
IR14
IP2
IP9
IP12
IP14
IA1
IA2
IA3
IA4
IA10
IA12
IA13
IA15
IR3
IR5
IR6
IR8
IR12
IP3
IP5
IP6
IP7
IP8
IP10
IP11
IP16
IA8
IA11
IA16
IA17
IA18
IR1
IR11
Difference between
answered and actual
Difference
between answered
and actual
L&C (km/hr) SLC Distance (m)
-10
-5.11
1
-4.3
-8
59.21
-2
75.03
-9
-4.77
-7
-0.64
0
6.61
-2
5.12
0
51.09
-13
-35.19
-5
10.65
4
-9.91
-1
-0.45
0
17.68
3
7.12
-5
8.16
0
-1.65
-7
-13.45
-5
46.3
-5
31.82
-4
-7.77
0
38.57
-1
-7.39
-1
0.44
0
1.87
7
2.96
-3
30.52
11
19.32
-1
17.71
-11
-7.49
0
9.1
0
24
-5
-6.54
1
-5.83
-5
-4.56
-8
6.17
-5
-10.91
-4
-4.79
0
-1.15
-14
-0.49
-8
6.16
8
15.64
-4
5.03
Difference between
answered and actual
Rock (RPM) Gravel (RPM) SLC (km/hr)
138
-15
1
116
139
-1
-93
-59
-3
-146
-18
2
-16
-117
-5
180
254
0
241
-13
0
587
158
-1
161
94
-4
463
242
-9
164
220
-1
-111
-152
-3
144
-16
-6
229
227
0
295
-83
3
210
-19
-2
176
211
-1
243
-25
-1
337
-14
-1
-36
-109
-4
634
484
2
210
219
1
227
106
-6
390
337
1
-141
-101
0
216
-252
2
372
68
2
213
251
-4
-106
-54
-4
61
-53
-4
62
-151
-7
41
-256
-5
44
18
-2
118
45
-4
48
85
0
262
164
2
-202
-204
0
495
-89
-1
-448
-591
-7
99
-70
-7
140
-285
-3
902
610
-1
438
208
0
Clusters
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
2
113
F.E Rating
(Rock)
3
2
3
2
3
1
1
2
3
3
2
1
2
3
4
2
2
1
2
2
1
1
3
1
1
2
4
1
3
3
2
2
3
3
2
2
3
2
1
1
3
3
1
Operator Id
IP1
IP2
IP3
IP5
IP6
IP8
IP10
IP11
IP14
IP15
IP16
IA4
IA5
IA6
IA10
IA16
IA17
IR5
IP9
IA3
IA8
IA18
IR1
IR6
IR7
IR8
IR10
IR14
IP7
IP12
IP13
IA1
IA2
IA11
IA12
IA13
IA14
IA15
IR2
IR3
IR9
IR11
IR12
Lifting Force
Breakout Force
4
3
3
4
3
3
3
3
4
4
3
3
3
3
4
3
3
4
3
3
3
3
3
3
1
3
4
2
3
3
3
3
3
4
3
3
4
3
2
3
3
3
2
3
3
3
3
4
4
3
4
3
3
4
3
3
3
4
4
3
3
3
3
4
3
4
3
3
4
4
4
3
4
3
4
3
4
4
4
4
3
4
3
3
4
3
2
4
3
3
4
3
3
3
2
2
4
3
2
3
3
2
3
4
3
2
2
2
4
3
4
4
4
4
3
4
3
4
3
4
4
4
4
3
4
3
3
4
3
2
3
3
4
3
3
3
3
3
3
3
3
3
4
4
3
4
4
4
3
2
3
4
4
4
4
4
4
3
4
3
4
4
4
4
4
4
3
4
4
3
4
3
2
4
2
3
4
2
3
3
3
3
4
3
3
4
3
3
4
3
4
2
3
2
3
3
4
3
3
4
3
4
4
4
2
3
3
3
4
3
3
3
3
4
3
2
4
3
3
4
3
3
4
3
3
4
2
3
4
4
3
4
4
4
3
4
4
4
4
4
4
3
4
4
4
4
4
4
4
4
4
4
3
4
4
3
4
3
2
2
3
3
2
2
2
1
2
2
2
3
3
2
4
3
4
3
3
2
2
2
3
2
4
3
3
4
3
3
2
3
4
3
3
4
2
3
4
2
3
3
3
3
3
3
3
3
2
3
3
4
3
2
2
3
3
4
3
4
4
3
3
3
3
3
3
4
4
4
4
3
2
3
3
4
3
3
4
3
3
4
3
3
4
3
Lifting Force Lifting Force Breakout Breakout Power (Gradeability) Overall Power
F.E Rating Rim pull Rating
Rating
Rating
Rating
Rating Force Rating Force Rating
(Gravel)
(Rock)
(Gravel)
(Rock)
(Gravel)
Fuel Efficiency Rating
4
2
3
3
4
4
2
4
3
2
4
3
3
4
3
3
3
4
3
3
2
4
4
4
4
4
3
4
3
3
2
3
4
4
4
3
4
2
4
3
4
3
4
Sound
4
3
3
3
3
4
3
4
3
2
4
3
3
4
2
3
4
4
3
3
2
4
3
4
4
3
3
4
3
3
4
3
3
4
3
3
4
2
4
3
3
3
4
Vibrations
4
3
3
3
3
3
3
3
3
3
4
3
3
3
2
3
4
3
3
3
3
4
4
4
3
4
2
4
4
3
4
3
3
4
2
2
4
3
3
3
4
3
3
4
3
3
3
3
4
3
3
3
2
3
2
3
3
3
3
4
4
3
3
2
4
3
4
4
3
3
4
4
3
3
2
3
4
3
3
4
3
3
3
4
3
3
Visibility Comfort level
CAB Experience
4
4
3
3
4
4
3
4
4
3
3
2
3
2
3
3
4
4
3
3
3
4
4
4
0
4
0
4
4
3
4
4
3
4
3
4
4
3
4
3
3
4
4
BSS
3
3
4
3
4
3
3
4
3
3
4
4
3
4
3
3
3
3
3
2
2
4
4
3
3
3
4
4
3
3
2
2
3
3
4
3
2
3
2
2
4
3
4
3
3
3
3
3
3
3
4
3
2
4
3
3
3
4
3
4
4
3
3
2
3
3
3
4
4
3
4
3
3
4
4
3
4
3
3
3
3
4
3
4
3
3
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
Gear Shift Overall feel of
Clusters
Jerk
L220
6.2 Clustering of preferences from chapter II
114
IP1
IP2
IP5
IP9
IP11
IP15
IP16
IA4
IA5
IA6
IA10
IA16
IA17
IR1
IR10
IR14
IP3
IP6
IP8
IP10
IP14
IA3
IA8
IA18
IR5
IR6
IR7
IR8
IP7
IP12
IP13
IA1
IA2
IA11
IA12
IA13
IA14
IA15
IR2
IR3
IR9
IR11
IR12
Operator Id
Fuel Efficiency
Rim pull Rating (20%)
Rock Gravel
90
120
60
60
90
60
60
120
60
60
90
60
60
90
80
90
120
60
60
90
80
30
90
60
60
90
60
90
90
60
120
120
80
60
90
80
60
90
60
90
90
80
120
120
80
30
60
80
90
90
60
90
90
80
30
90
80
30
90
60
90
120
60
60
90
60
30
90
80
30
90
60
30
120
60
30
90
60
30
30
60
60
90
80
90
90
60
90
90
80
60
90
60
60
90
80
90
90
60
90
120
80
60
90
80
60
90
80
90
120
80
60
90
60
30
60
80
30
90
60
90
90
60
90
90
80
30
60
60
Lifting Force
Rock Gravel
30
30
60
45
45
60
45
60
45
45
30
45
60
45
45
45
30
45
45
60
45
60
30
45
45
60
60
60
60
60
60
60
45
45
60
45
45
45
45
45
30
45
30
45
30
30
30
45
60
60
45
60
60
60
60
60
45
45
60
60
45
45
60
60
45
60
60
60
60
60
60
60
60
60
45
45
60
60
45
60
45
45
60
60
45
45
Breakout Force (10%)
Rock
Gravel
20
20
40
40
30
30
40
40
30
40
30
30
40
40
30
20
30
30
40
40
30
40
30
30
40
40
30
40
30
30
40
40
20
30
40
40
20
30
30
30
30
30
20
30
30
40
20
40
30
40
30
40
40
40
30
40
30
40
40
40
40
40
40
40
20
40
30
40
30
40
30
40
40
40
30
30
30
40
30
40
30
30
40
40
30
30
30
30
45
45
15
30
30
45
45
30
60
45
60
45
45
60
45
30
30
30
30
30
30
30
45
30
60
45
45
45
30
45
60
45
45
60
30
45
60
30
45
45
45
30
30
30
30
30
30
20
20
30
30
40
30
40
30
40
40
30
30
20
30
40
30
30
30
40
30
40
40
30
20
30
30
40
30
30
40
30
30
40
30
30
40
30
430
455
480
470
435
465
465
385
420
485
595
440
495
525
585
470
455
505
390
390
475
395
390
375
485
415
420
505
475
525
440
505
505
555
495
520
550
435
460
415
465
545
375
Power (Gradeability) (15%) Overall Power (10%) Total Weighted Score
Clusters
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
6.3 Clustering of parameters from chapter II using weighted scores
115
IP1
IP2
IP3
IP5
IP6
IP7
IP8
IP9
IP10
IP11
IP12
IP13
IP14
IP15
IP16
IA1
IA2
IA3
IA4
IA6
IA8
IA10
IA11
IA12
IA15
IA17
IA5
IA13
IA14
IA16
IA18
IR1
IR2
IR3
IR6
IR8
IR9
IR10
IR12
IR5
IR7
IR11
IR14
Operator ID
Fuel Efficiency
SLC (Gravel) L&C (Gravel)
28.568
12.383
28.795
11.232
23.194
11.033
25.077
11.073
27.940
11.562
27.126
12.592
27.992
9.829
22.729
10.324
24.499
11.003
26.780
10.967
28.455
11.584
23.807
11.393
30.685
12.133
28.800
12.416
28.974
11.618
27.991
9.346
29.040
11.531
29.764
8.705
25.649
10.890
24.128
9.884
29.624
11.814
30.368
10.674
29.592
12.012
28.633
10.901
26.785
11.365
23.325
9.847
28.322
9.929
23.422
11.316
26.376
11.301
18.006
9.659
25.247
9.961
22.488
10.704
26.134
9.473
24.634
10.640
27.278
8.766
22.429
12.012
25.526
11.875
18.415
5.219
22.748
7.117
27.415
11.559
19.368
10.748
26.529
10.131
22.572
8.200
27.400
11.131
23.969
10.054
23.727
9.551
Productivity
SLC (Rock) SLC (Gravel) L&C (Gravel)
16.982
820.918
383.975
21.864
995.960
383.896
14.990
728.105
339.954
17.956
750.692
368.652
20.013
771.708
361.543
20.933
766.830
362.546
19.924
710.726
348.481
13.416
608.365
366.576
17.970
740.501
355.756
18.533
654.606
301.093
20.462
792.458
323.688
17.892
745.572
391.875
19.091
941.888
351.025
20.083
795.892
388.524
21.074
789.442
300.341
16.692
800.475
368.736
17.477
721.533
338.694
19.375
886.519
372.451
13.313
665.885
267.844
13.847
832.146
388.053
17.843
767.114
376.845
18.698
614.402
260.902
14.767
867.519
369.055
12.331
547.096
296.509
16.010
648.299
280.214
11.067
636.420
309.417
18.591
746.450
326.570
12.941
530.489
273.772
15.686
681.595
353.687
12.121
616.186
355.577
15.716
679.566
330.543
16.448
354.898
233.624
9.446
557.128
242.357
7.584
379.864
257.248
11.105
499.447
150.189
12.233
515.578
369.055
16.660
466.206
279.373
7.521
282.529
109.850
12.338
386.908
150.814
13.214
632.614
283.960
13.118
420.224
219.548
11.602
532.509
262.618
10.991
243.894
162.853
17.662
758.586
345.889
12.855
534.335
276.251
12.253
443.230
215.959
SLC (Rock)
521.074
784.206
483.412
554.793
452.264
536.878
432.139
395.577
551.838
500.215
581.171
558.213
526.327
567.641
523.780
425.330
477.820
576.044
253.737
532.687
440.898
412.260
450.741
200.177
389.114
270.127
507.110
323.851
417.748
429.634
349.582
287.081
219.044
106.531
192.061
243.816
309.387
126.453
236.213
353.255
230.509
216.424
166.880
485.133
290.956
240.656
Output Torque (Overall Power)
Rimpull (kN) Lifting Force (Mpa)
Gradeability
SLC (Rock)
SLC (Rock)
SLC (Gravel) Median L&C (Gravel) Median SLC (Rock) Median Average Output Torque
201.7053602
26.17819405
1329
1437
1329
1427.528
281.4590981
27.30203056
1437
1600
1464
1303.109
103.7384833
27.44808388
1383
1383
1356
1063.782
217.8869196
27.34985352
1437
1519
1356
1291.605
230.5354891
26.08125496
1437
1491
841
1369.201
188.534738
25.78785706
1464
1491
1112
1219.379
219.7345462
28.03423691
1139
1383
1058
1181.865
219.1458667
27.83906746
1410
1437
1139
1380.104
210.9674393
26.42183113
1342
1437
1193
1267.358
272.8771961
28.03423691
1193
1410
1329
1196.44
256.7412687
27.00927734
1166
1383
922
1298.041
208.7487058
25.2992878
1464
1519
1410
1333.533
200.9648766
28.52280617
1464
1546
1220
1216.974
108.0237973
27.49720001
1383
1410
1383
1340.02
217.391417
26.86193085
1220
1193
1227
1167.46
281.8017236
25.93391037
1383
1437
1085
1542.766
208.8921976
23.93181229
1329
1329
1302
1251.636
288.6099602
27.7408371
1437
1546
1437
1277.824
205.2183905
26.32489395
1274
976
813
1256.721
252.3845911
27.88818359
1464
1464
1437
893.528
189.433672
23.63905907
1112
1247
1166
1450.606
192.8254745
26.76370049
1030
1193
1139
1170.372
195.4211316
20.90410995
1220
1491
1139
1334.691
233.1554637
25.20105743
732
1166
759
1185.451
276.9502112
25.93391037
1085
1437
813
1190.561
209.3819287
25.73874092
1261
1085
1139
1159.035
222.2233424
26.91104698
1396
1519
1274
1260.293
202.6535588
26.81281662
1085
1112
1003
1229.739
218.1206109
26.76370049
1464
1410
1329
1449.961
293.8384329
27.78995323
1356
1437
1410
1346.61
251.7863904
25.88479424
1329
1247
786
1184.915
284.5152599
28.71797371
651
1085
542
1312.275
215.5552034
26.61829376
1193
1112
1030
1227.729
238.1585324
12.74708462
732
766
678
1188.228
242.7996165
26.5200634
840
705
542
769.393
240.5566759
26.27577782
1274
1491
841
1334.691
103.1909506
26.61829376
813
1058
759
1135.186
214.4532192
27.83906746
624
678
515
1076.622
230.2272379
26.47094727
976
651
1058
1050.369
252.6392967
26.56917763
1274
1030
1302
1041.689
216.8072435
26.37400818
1193
990
678
999.044
223.2272922
22.27093887
1003
1030
813
1029.09
232.9106585
10.79410362
434
651
569
1090.562
218.526
26.397
1293.360
1397.000
1177.160
1264.422
225.941
25.788
1078.308
1131.154
911.385
1205.744
231.162
22.496
976.000
870.400
884.000
1042.151
Average Bucket Loads
SLC (Gravel) L&C (Gravel) SLC (Rock)
9792.308519 9072.229709 8114.276542
9241.676351 7849.977173 7190.007878
9530.006836 9245.836589 8304.162882
8790.930309 8409.103871 8040.642456
9192.317082 8690.907759 7413.658958
9233.308702 9260.009766 8245.798014
8727.281605 8574.971826 7222.734686
9314.973828 9027.772732 6291.685649
8434.993115 8800.032043 7267.867676
7440.867676 7600.000577 7034.622183
7822.238715 7949.987061 8565.369291
8577.29177 8970.835083 8522.213162
8833.330946 8299.982352 7890.030827
9173.078576 8964.285854 9325.002848
8409.103915 7920.008984 8003.57659
8054.163249 8029.985303 7245.840169
8740.918945 8285.71861 7788.902778
8981.810369 7710.719448 8353.82035
9154.149251 7978.597726 4969.961735
8534.628718 8300.012634 7299.984172
8477.269132 9027.283203 7169.998193
7742.289476 7719.976953 7010.70445
8979.1757 8781.809304 6207.664776
7923.097957 8175.02915 3913.661488
8359.070623 8133.329549 6284.637395
9396.148212 8716.675252 5783.345835
8864.283901 8133.329427 8833.344401
8625.02832 7981.804776 5846.702995
8194.998242 8604.544567 6913.678178
8763.655451 8890.908736 7699.999642
9282.156738 8865.010986 6814.266166
7063.601851 8774.949158 7437.515177
8013.352344 6810.018213 4763.655257
7809.122647 7366.689738 2956.234451
7721.417236 6435.722517 4186.366255
8050.031062 8781.809304 5750.003153
8633.328532 9164.986084 6866.670492
5189.987158 5181.238464 3295.01521
7045.844564 6268.733521 5804.167358
8538.468938 9392.856585 6633.352356
8795.446644 8318.764221 5622.725874
7014.287772 7384.98291 4603.573486
4379.155752 5837.510071 3370.816854
8698.411
8431.136
7347.073
8123.624
7977.514
5934.369
7154.641
7440.569
5206.927
Clusters
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
Cluster 1 average
Cluster 2 average
Cluster 3 average
6.4 Clustering of actual values of parameters from chapter II
116
IP1
IP2
IP3
IP5
IP6
IP8
IP9
IP10
IP11
IP14
IP15
IP16
IA3
IA4
IA5
IA6
IA8
IA10
IA16
IA17
IR1
IP7
IP12
IP13
IA1
IA2
IA11
IA12
IA13
IA14
IA15
IR2
IR9
IR12
IA18
IR3
IR5
IR6
IR7
IR8
IR10
IR11
IR14
Operator Id
Gravel
3
4
2
3
3
3
2
4
3
3
1
3
2
3
1
3
2
3
3
4
3
4
2
3
2
3
1
3
2
3
3
3
1
3
4
4
2
3
2
3
3
3
3
3
3
3
2
3
2
3
3
3
3
4
2
3
2
3
3
4
2
3
1
2
3
3
1
2
1
3
1
3
1
4
1
3
1
1
2
3
4
4
3
3
1
2
2.2
3.25
2.46153846 3.07692308
1.6
2.8
Rock
Fuel Efficiency Rating
3
3
3
3
4
4
3
3
4
3
3
4
3
3
3
3
4
4
4
3
4
3
4
3
4
3
4
4
4
4
3
4
3
3
3
3
3
3
3
4
4
4
4
3.35
3.615384615
3.4
Rim pull Rating
Gravel
Rock
Gravel
Breakout Force
2
2
2
2
4
3
4
4
3
3
2
3
3
4
3
3
4
3
4
4
3
3
2
3
3
4
4
4
3
3
3
3
3
3
3
4
2
3
3
3
2
3
3
3
4
3
4
4
2
3
2
3
3
3
3
2
2
3
3
3
3
4
4
4
2
2
3
4
3
4
3
4
2
3
3
3
3
4
4
4
4
4
3
4
3
3
3
4
4
4
4
4
3
3
4
4
4
4
4
4
3
4
2
4
4
4
3
4
4
4
3
4
4
4
3
4
4
4
4
4
3
3
3
3
4
4
3
4
3
3
3
3
3
3
3
3
2
3
2
4
3
4
3
4
4
4
3
4
3
4
3
4
4
4
4
4
4
4
3
4
4
4
3
3
4
4
4
4
4
4
4
4
2.8
3.15
3.1
3.35
3.61538462 3.69230769 3.23076923 3.84615385
3.5
3.8
3.2
3.8
Rock
Lifting Force
2
2
3
3
2
2
3
2
1
2
2
2
2
3
3
2
2
4
3
4
3
3
3
2
3
4
3
3
4
2
3
4
3
3
2
2
3
2
4
3
3
3
4
2.45
3.076923077
2.9
3
3
3
3
3
2
3
3
3
4
3
2
3
2
3
3
3
4
3
4
3
3
2
3
3
4
3
3
4
3
3
4
3
3
3
3
4
3
4
4
4
4
4
3
3.153846154
3.6
Power (Gradeability) Overall Power
4
3
3
3
3
4
3
3
4
3
2
4
3
3
3
4
2
2
3
4
3
3
3
4
3
3
4
3
3
4
2
4
3
4
4
3
4
4
4
3
3
3
4
Vibrations
4
3
3
3
3
3
3
3
3
3
3
4
3
3
3
3
3
2
3
4
4
4
3
4
3
3
4
2
2
4
3
3
4
3
4
3
3
4
3
4
2
3
4
Visibility
4
3
3
3
3
4
3
3
3
3
2
3
3
2
3
3
2
3
3
4
3
4
3
3
2
3
4
3
3
4
3
3
4
3
4
3
4
4
4
3
3
3
4
Comfort
level
4
4
3
3
4
4
3
3
4
4
3
3
3
2
3
2
3
3
3
4
4
4
3
4
4
3
4
3
4
4
3
4
3
4
4
3
4
4
0
4
0
4
4
BSS
3
3
4
3
4
3
3
3
4
3
3
4
2
4
3
4
2
3
3
3
4
3
3
2
2
3
3
4
3
2
3
2
4
4
4
2
3
3
3
3
4
3
4
3
3
3
3
3
3
3
3
4
3
2
4
3
3
3
3
2
4
3
4
3
3
3
4
4
3
4
3
3
3
3
4
4
3
3
3
4
3
4
4
3
3
4
Gear Shift Overall feel
Jerk
of L220
3.1
3.15
3.1
3
3.25
3.2
3.1
3.38461538 3.23076923 3.30769231 3.23076923 3.61538462 2.92307692 3.38461538
3.7
3.6
3.3
3.5
3.1
3.3
3.4
4
2
3
3
4
4
3
2
4
3
2
4
3
3
3
4
2
3
3
3
4
3
3
2
3
4
4
4
3
4
2
4
4
4
4
3
4
4
4
4
3
3
4
Sound
CAB Experience
6.5 Clustering of combination of preferences and actual values from chapter II
117
118
27.01835709
25.810177
23.17459631
26.56731167 11.18407716 15.9803249 652.3152425 316.4714433 398.196473 221.0985403
23.66357987 9.435257612 11.5423008 457.3132347 229.6676485 222.172456 234.3566163
967.9
1157.615385
1325.2
923.9
1310
1385.65
778.2
1015.769231
1235.5
Average Bucket Loads
SLC (Gravel) L&C (Gravel) SLC (Rock) Clusters
9792.308519 9072.229709 8114.276542
1
9241.676351 7849.977173 7190.007878
1
9530.006836 9245.836589 8304.162882
1
8790.930309 8409.103871 8040.642456
1
9192.317082 8690.907759 7413.658958
1
8727.281605 8574.971826 7222.734686
1
9314.973828 9027.772732 6291.685649
1
8434.993115 8800.032043 7267.867676
1
7440.867676 7600.000577 7034.622183
1
8833.330946 8299.982352 7890.030827
1
9173.078576 8964.285854 9325.002848
1
8409.103915 7920.008984 8003.57659
1
8981.810369 7710.719448 8353.82035
1
9154.149251 7978.597726 4969.961735
1
8864.283901 8133.329427 8833.344401
1
8534.628718 8300.012634 7299.984172
1
8477.269132 9027.283203 7169.998193
1
7742.289476 7719.976953 7010.70445
1
8763.655451 8890.908736 7699.999642
1
9396.148212 8716.675252 5783.345835
1
7063.601851 8774.949158 7437.515177
2
9233.308702 9260.009766 8245.798014
2
7822.238715 7949.987061 8565.369291
2
8577.29177 8970.835083 8522.213162
2
8054.163249 8029.985303 7245.840169
2
8740.918945 8285.71861 7788.902778
2
8979.1757 8781.809304 6207.664776
2
7923.097957 8175.02915 3913.661488
2
8625.02832 7981.804776 5846.702995
2
8194.998242 8604.544567 6913.678178
2
8359.070623 8133.329549 6284.637395
2
8013.352344 6810.018213 4763.655257
2
8633.328532 9164.986084 6866.670492
2
7045.844564 6268.733521 5804.167358
3
9282.156738 8865.010986 6814.266166
3
7809.122647 7366.689738 2956.234451
3
8538.468938 9392.856585 6633.352356
3
7721.417236 6435.722517 4186.366255
3
8795.446644 8318.764221 5622.725874
3
8050.031062 8781.809304 5750.003153
3
5189.987158 5181.238464 3295.01521
3
7014.287772 7384.98291 4603.573486
3
4379.155752 5837.510071 3370.816854
3
Cluster 1
1251.02175 8839.755163 8446.630642 7460.971398
average
Cluster 2
1285.457576 8324.582689 8378.692817 6815.562244
average
Cluster 3
1076.4603 7382.591851 7383.331832 4903.652116
average
Output Torque (Overall Power)
Rimpull (kN) Lifting Force (Mpa)
Gradeability
SLC (Rock)
SLC (Rock)
SLC (Gravel) Median L&C (Gravel) Median SLC (Rock) Median Average Output Torque
201.7053602
26.17819405
1329
1437
1329
1427.528
281.4590981
27.30203056
1437
1600
1464
1303.109
103.7384833
27.44808388
1383
1383
1356
1063.782
217.8869196
27.34985352
1437
1519
1356
1291.605
230.5354891
26.08125496
1437
1491
841
1369.201
219.7345462
28.03423691
1139
1383
1058
1181.865
219.1458667
27.83906746
1410
1437
1139
1380.104
210.9674393
26.42183113
1342
1437
1193
1267.358
272.8771961
28.03423691
1193
1410
1329
1196.44
200.9648766
28.52280617
1464
1546
1220
1216.974
108.0237973
27.49720001
1383
1410
1383
1340.02
217.391417
26.86193085
1220
1193
1227
1167.46
288.6099602
27.7408371
1437
1546
1437
1277.824
205.2183905
26.32489395
1274
976
813
1256.721
222.2233424
26.91104698
1396
1519
1274
1260.293
252.3845911
27.88818359
1464
1464
1437
893.528
189.433672
23.63905907
1112
1247
1166
1450.606
192.8254745
26.76370049
1030
1193
1139
1170.372
293.8384329
27.78995323
1356
1437
1410
1346.61
209.3819287
25.73874092
1261
1085
1139
1159.035
284.5152599
28.71797371
651
1085
542
1312.275
188.534738
25.78785706
1464
1491
1112
1219.379
256.7412687
27.00927734
1166
1383
922
1298.041
208.7487058
25.2992878
1464
1519
1410
1333.533
281.8017236
25.93391037
1383
1437
1085
1542.766
208.8921976
23.93181229
1329
1329
1302
1251.636
195.4211316
20.90410995
1220
1491
1139
1334.691
233.1554637
25.20105743
732
1166
759
1185.451
202.6535588
26.81281662
1085
1112
1003
1229.739
218.1206109
26.76370049
1464
1410
1329
1449.961
276.9502112
25.93391037
1085
1437
813
1190.561
215.5552034
26.61829376
1193
1112
1030
1227.729
103.1909506
26.61829376
813
1058
759
1135.186
230.2272379
26.47094727
976
651
1058
1050.369
251.7863904
25.88479424
1329
1247
786
1184.915
238.1585324
12.74708462
732
766
678
1188.228
252.6392967
26.56917763
1274
1030
1302
1041.689
242.7996165
26.5200634
840
705
542
769.393
216.8072435
26.37400818
1193
990
678
999.044
240.5566759
26.27577782
1274
1491
841
1334.691
214.4532192
27.83906746
624
678
515
1076.622
223.2272922
22.27093887
1003
1030
813
1029.09
232.9106585
10.79410362
434
651
569
1090.562
485.78822 216.9173141
26.66091794 10.84874956 17.3376457 753.6962381 345.3736815
Fuel Efficiency
Productivity
SLC (Gravel) L&C (Gravel) SLC (Rock) SLC (Gravel) L&C (Gravel) SLC (Rock)
28.568
12.383
16.982
820.918
383.975
521.074
28.795
11.232
21.864
995.960
383.896
784.206
23.194
11.033
14.990
728.105
339.954
483.412
25.077
11.073
17.956
750.692
368.652
554.793
27.940
11.562
20.013
771.708
361.543
452.264
27.992
9.829
19.924
710.726
348.481
432.139
22.729
10.324
13.416
608.365
366.576
395.577
24.499
11.003
17.970
740.501
355.756
551.838
26.780
10.967
18.533
654.606
301.093
500.215
30.685
12.133
19.091
941.888
351.025
526.327
28.800
12.416
20.083
795.892
388.524
567.641
28.974
11.618
21.074
789.442
300.341
523.780
29.764
8.705
19.375
886.519
372.451
576.044
25.649
10.890
13.313
665.885
267.844
253.737
28.322
9.929
18.591
746.450
326.570
507.110
24.128
9.884
13.847
832.146
388.053
532.687
29.624
11.814
17.843
767.114
376.845
440.898
30.368
10.674
18.698
614.402
260.902
412.260
18.006
9.659
12.121
616.186
355.577
429.634
23.325
9.847
11.067
636.420
309.417
270.127
22.488
10.704
16.448
354.898
233.624
287.081
27.126
12.592
20.933
766.830
362.546
536.878
28.455
11.584
20.462
792.458
323.688
581.171
23.807
11.393
17.892
745.572
391.875
558.213
27.991
9.346
16.692
800.475
368.736
425.330
29.040
11.531
17.477
721.533
338.694
477.820
29.592
12.012
14.767
867.519
369.055
450.741
28.633
10.901
12.331
547.096
296.509
200.177
23.422
11.316
12.941
530.489
273.772
323.851
26.376
11.301
15.686
681.595
353.687
417.748
26.785
11.365
16.010
648.299
280.214
389.114
26.134
9.473
9.446
557.128
242.357
219.044
25.526
11.875
16.660
466.206
279.373
309.387
22.748
7.117
12.338
386.908
150.814
236.213
25.247
9.961
15.716
679.566
330.543
349.582
24.634
10.640
7.584
379.864
257.248
106.531
27.415
11.559
13.214
632.614
283.960
353.255
27.278
8.766
11.105
499.447
150.189
192.061
19.368
10.748
13.118
420.224
219.548
230.509
22.429
12.012
12.233
515.578
369.055
243.816
18.415
5.219
7.521
282.529
109.850
126.453
26.529
10.131
11.602
532.509
262.618
216.424
22.572
8.200
10.991
243.894
162.853
166.880
Fly UP