Automatic Time Study Method for Recording Work Phase Times of

Timber Harvesting
Teijo Palander, Yrjo Nuutinen, Arto Kariniemi, and Kari Vaatainen
Abstract: The objective of the present study was to develop an automatic time study method based on a
process-data model for single-grip harvesters, with inputs based on data automatically collected by the
harvesters onboard computer. The method integrates the phases of the work cycle into components under
conditions in which the work phases may overlap to varying degrees. During the work phase analysis, we found
that process-data models differed under similar work conditions because work phases could not be completely
separated during the automatic recording of data. We therefore used the combined data provided by manual and
automatic timing to develop a new process-data model for a single-grip harvesters work. We also analyzed the
overlapping and simultaneous work phases to optimize the improved process-data model. The results were
satisfactory, and the method can be systematically used in time studies based on automatically recorded data by
modifying the process-data model using the approach described in this article. Adjustment of the model to
improve data-recording accuracy compared with manual time studies has great potential, but this must be
confirmed through additional harvesting experiments during work studies with different machines and in
different forests. FOR. SCI. 59(4):472 483.
Keywords: time and motion studies, single-grip harvester, process data, data management system

INCE THE INTRODUCTION of the first single-grip harvesters (hereafter, harvesters) in the Nordic countries in the 1970s, studies of the harvesters work
cycle have relied on the time study method. Since then,
these studies have expanded from the testing of new models
to determining the influence of the operating environment,
the operational efficiency of the integration of harvesting
with downstream processes such as forwarding harvesting
chains, operator skills, and the dynamics of human-machine
systems. Research methodologies have evolved greatly
since the introduction of these machines (Figure 1).
In the 1970s and 1980s, time studies were mainly conducted using digital watches (International Labour Office
1981). In the mid-1980s, field computers started to replace
digital watches and paper forms in time studies because they
provided opportunities for measuring the time elements of a
work cycle in more detail and more accurately (Harstela
1988). During the 1990s, numerous time studies of harvesters were conducted using handheld field computers (e.g.,
Kellogg and Bettinger 1994, Eliasson 1998), and these
devices remained essential into the 2000s (Karha et al.
2004, Puttock et al. 2005, Kariniemi 2006, Spinelli and
Visser 2008, Ovaskainen 2009). Since the 1990s, digital
video cameras have been used to record harvester performance and working techniques (Vaatainen et al. 2005,
Ovaskainen et al. 2006, Nurminen et al. 2006, Nakagawa et
al. 2007). In the 2000s, it became possible to collect time
study data automatically using a harvesters computer connected to channels such as the controller-area network

(CAN) hardware (Vaatainen et al. 2005, Kariniemi 2006,

Tikkanen et al. 2008, Ovaskainen 2009, Nuutinen et al.
2010, Palander et al. 2012). The automated time study
methods used for monitoring the performance of harvesters
in cut-to-length systems have also been used to monitor
tree-length harvesting systems (McDonald and Fulton
2005).
The CAN bus was developed and launched by Robert
Bosch Corporation in 1986. It was designed specifically for
automotive applications and is a multiplexed wiring system
that can be used to connect intelligent devices such as
electronic control units in vehicles, allowing data to be
transferred in a low-cost and reliable manner (CAN in
Automation 2011). The benefit of the CAN bus for time
studies of harvester work is that it lets researchers record
large amounts of time study data with a high level of detail
and accuracy during the harvesters work on each processed
stem (Figure 2). A variety of dataloggers have been developed that can automatically record this information, including the Forestry Engineering Research Institute of Canadas
MultiDAT (FPInnovations 2012) and the PlusCan datalogger (Plustech Oy, Tampere, Finland).
For harvesters, Plustech Oy developed a datalogger to
automatically record the information flow in the CAN bus
channels. Their first device for recording the CAN bus
information (the PlusCan datalogger) recorded detailed information concerning the harvesters operations, such as the
stem dimensions and time consumption during each of
the harvesters work and movement phases (Peltola 2003).

Manuscript received February 1, 2012; accepted September 1, 2012; published online October 4, 2012; http://dx.doi.org/10.5849/forsci.12-009.
Teijo Palander (teijo.s.palander@uef.fi), University of Eastern Finland, Joensuu, Finland. Yrjo Nuutinen (yrjo.nuutinen@metla.fi), Finnish Forest Research
Institute. Arto Kariniemi (arto.kariniemi@metsateho.fi), Metsateho Oy. Kari Vaatainen (kari.vaatainen@metla.fi), Finnish Forest Research Institute.
Acknowledgments: The research reported in this article was partially funded by the Finnish Funding Agency for Technology and Innovation (decision number
40197/09) and Metsateho Oy.
Copyright 2013 by the Society of American Foresters.

472

Forest Science 59(4) 2013

Figure 1.

Figure 2.

The development of time studies in harvester operations.

The ideal work cycle of a single-grip harvester.

The successor to the PlusCan datalogger, the TimberLink

(developed by John Deere), is a more advanced monitoring

system that has been available as an option on all new John

Deere harvesters since 2005. TimberLink comprises hardware and software that collect and process CAN bus data
about the machines and the operators condition and performance (John Deere 2008, Tikkanen et al. 2008, Nuutinen
et al. 2010, Palmroth 2011, Palander et al. 2012).
The first Forest Work Study Nomenclature (Nordic Forest Study Council 1978) was published in 1963 and revised
in 1978. It represented an agreement between Denmark,
Finland, Sweden, and Norway about terminology and methodology that was worked out by the Nordic Work Study
Council. The Nomenclature aimed to improve the comparability of international time study reports (Samset 1990).
The second international Forest Work Study Nomenclature
(Bjorheden 1991) was launched in 1995 (International
Union of Forestry Research Organisations 1995) for practical testing aiming at refining it for final acceptance at the
International Union of Forestry Research Organisations
World Congress in Malaysia in 2000. These nomenclatures
were the first steps toward developing a common universal
time study methodology. They contained a collective proposal for the basic time concepts and phases that should be
used for time measurements in forestry work to serve as a
basis for any study that hoped to achieve international
significance.
Spinelli et al. (2010) have proposed that general models
should be developed for harvesters and processors instead
Forest Science 59(4) 2013

473

of using more accurate stand-level models for a separate

human-machine system. The goal of such a generalized
model would be to eliminate the time-consuming and expensive process of creating a specific model for each harvesting machine and stand. Recently, we developed an
adaptive work study method to support stand-level studies
(Palander et al. 2012). The method uses a detailed model of
a machines work cycle and work-phase data provided by
the machines automatic monitoring system to identify the
most important work phases in a human-machine system.
However, the only time study standard intended specifically
for forest machines is the Standard for Forest Data and
Communication (Skogforsk 2010), which is a de facto standard for all forest machines manufactured in the Nordic
countries.
The first version of StanForD was published in 1987 in
Sweden. In the early 1990s, Finnish researchers also joined
in the development of the StanForD standard (Arlinger et al.
2008). The goal of the standard was to enable analyses of
the technical and organizational factors that affect the operation of forwarders and harvesters. The latest version of
StanForD was issued in 2010 (Arlinger et al. 2010). In the
current version of StanForD for harvesters, the effective
work time is divided into processing and travel time at the
highest level. During StanForDs development process, information technology provided a number of possibilities to
promote the use of a standard method and overcome the
limitations of previous international standards for time studies. StanForD was specifically developed to account for the
heterogeneous timing devices that have been used, the research topics these techniques support, and the approaches
used in these studies (Figure 1).
Kariniemi and Vartiamaki (2010) developed a generalized model of the work phase classification for use in
automatic time studies of harvesters. This is a process-data
model in which pause times are considered in addition to the
effective work time. Despite the common consensus, the
protocols that have been implemented, and their continuous
use within the forest engineering community, there is still a
widespread misunderstanding of automatic time study approaches at the conceptual, theoretical, and practical levels
in the context of cut-to-length harvesting systems. The
process-data model of Kariniemi and Vartiamaki (2010) is
based on an idealized work cycle for the harvester wherein
the work phases follow regularly repeating steps (Figure 2).
However, unforeseen situations that lead to deviation from
the normal procedures can occur during the work, and this
can lead to difficulties in correctly identifying how the time
consumed by these activities should be allocated among the
harvesters work phases, especially when automatic rather
than manual recording of the time data is used. This is
problematic because it is important to ensure that all human
and machine actions are included in the right work phases
and that all work time is recorded as effective work. Unforeseen situations that can confuse the automatic recording
of effective time consumption include the following possibilities for each level 1 work phase in the model:
1.

474

Gripping the stem. Automatic recording can correctly

record the removal of undergrowth surrounding a tree
Forest Science 59(4) 2013

(clearing) within the work phase in which the operator swings the harvesters felling head toward the tree
and fells the undergrowth using the heads saw. However, undergrowth may also be removed by pressing
the felled tree against the vegetation or by dragging
the tree instead of using the harvesters felling saw. In
these cases, the operation is not registered correctly. A
prefelled tree can also cause confusion between the
gripping the stem phase and the felling phase because
the felling cut was made before the processing phase.
If the tree to be felled is sufficiently branchy, the butt
must be delimbed before the felling cut so that the
harvester head can grip the stem. This operation
should be included in the gripping the stem phase
because it is involved in the preparation for felling and
is not part of the felling.
2.

Felling. When the felling cut must be repeated several

times for a tree that is too large to fell in a single cut,
multiple felling cuts will be required, and these must
be combined into a single felling phase. It is also
possible to fell several small trees consecutively if the
felling head has an accumulator arm and then process
them simultaneously, which does not follow the ideal
work cycle shown in Figure 2. After the felling cut and
before stem feeding, damage to the butt of the tree
must be removed by sawing off a short piece of the
stem, and this operation should be included in the
processing work phase rather than the felling phase.

3.

Processing. Several operations can confuse the division of the recorded times during processing. The stem
can break during felling, in which case a small piece
must be cut from the first log to remove the damaged
part of the stem, and the activity should be included
under processing, not felling, despite the activation of
the heads felling saw. Furthermore, a tree that divides
into two stems near the stump must be separated into
two stems by a new felling cut. During cross-cutting,
the first cut might not suffice, leading to a requirement
for additional cuts. When the top of a stem or a whole
unmerchantable tree is fed through the felling head
and the diameter is too small to produce a merchantable log, the time should not be recorded as processing
because there is no output (defined as producing a
volume of merchantable wood). Sometimes stem feeding is conducted using only boom movements without
activating the feed rollers, and, in this case, the feeding time and the length of the stem cannot be recorded
by the automatic systems. The endpoint of processing
can also be defined in two ways: the final cross-cut of
the stem or return of the harvesters felling head to an
upright position. Which of the two definitions should
be used has not yet been defined in the existing time
study standards and is still being debated.

These and other reasons why automatic recording may

be difficult to use with a process-data model should be
clarified so that these problems can be accounted for during
model development and use. Another problem with automatic recording is the large amount of time study data

obtained for each processed stem, which must be systematically organized (i.e., divided among the work phases which
may overlap to varying degrees). The overall purpose of the
present study was to examine both automatic and manual
recording of the time consumption during a harvesters
work cycle to increase our understanding of the potential for
automatic recording during time studies and to identify
situations in which manual recording can provide insights
into better ways to allocate automatically recorded time
data. The study also aimed to identify the advantages of
automatic time studies.
We conducted a study in which the focus was to find
common characteristics, models, and new theoretical ideas,
methods, and concepts by analyzing two recording alternatives (automatic versus manual) under similar work conditions. In an experimental study strategy, tests are used as a
means of researching different phenomena. In our study, we
identified the key components that should be quantified
empirically (based on observations of the harvesters work)
instead of assuming from the start what those components
should be (Eisenhardt 1989, 1991, Dyer and Wilkins 1991).
We used the PlusCan datalogger for automatic recording
and visual observation by a researcher using a handheld
field computer for manual recording. We used the processdata model of Kariniemi and Vartiamaki (2010) as the basis
for the automatic recording, because this model had been
developed especially for use with automatic timing.
Our objective was to develop a systematic method for
recording details of the effective work time of a harvester
that would support the use of a process-data model. The fact
that harvesting machines operate so rapidly that accurate
manual time study is impossible, combined with the presence of many overlapping time elements, makes an automatic solution essential. The fact that the CAN bus data are
readily available permits such a solution. In this article, we
also describe the modifications required to the process-data
model to allow the use of the CAN bus data and especially
the allocation of all (overlapping) time elements within the
models hierarchical data structure. We used principal components analysis in the final stage of the work phase analysis to optimize the allocation of times among work phases.
One goal of our study was to develop a general model that
met the criteria of Spinelli et al. (2010) and would therefore
facilitate the use of automatically recorded data without
requiring researchers to develop a specific model for each
combination of machine and stand conditions.

Materials and Methods

The work phase analysis was performed in three stages.
In the first stage, we developed an original version of the
Kariniemi and Vartiamaki (2010) process-data model for
time studies of a harvesters work by automatically collecting time study data using the PlusCan datalogger and manual recordings taken using a Husky-Hunter handheld field
computer. In the second stage, we performed three field
tests (three different time studies) to obtain data that could
be used to improve the process-data model. Test 1 examined
the activities during the processing phase (automatic recording). Test 2 examined the activities that occur before the

processing work phase (manual recording). Test 3 examined

all three level 1 work phases (gripping the stem, felling, and
processing) in the model (automatic and manual recording).
In this second stage, the adjustment of the model was based
on automatic and manual time study data obtained in a
previous field study (Vaatainen et al. 2005). In the study of
Vaatainen et al., automatic timing was conducted using a
PlusCan datalogger as a recording device. Manual timing
was conducted by an observer using a Rufco handheld field
computer.
In the third stage, we developed an automatic time study
methodology by combining the manual and automatic timing data to develop an improved process-data model of the
harvesters work. Test 3 can be used to examine all three
level 1 work phases (gripping the stem, felling, and processing) in the model and to develop the general model. An
experimental research strategy was applied for identifying
the components of the harvesters work cycle, and these
results were used to modify the model. Specifically, we
performed an exploratory factor analysis to compress the
durations of the work phases and the subphases of the
harvesters work into the most significant factors and to
identify any latent structures in the processing phase.

Process-Data Model
The process-data model of Kariniemi and Vartiamaki
(2010) was developed specifically to use the data provided
by the CAN bus of a harvester; in this study, we refer to this
as the process data. Process data include detailed information about harvester operations such as the stem dimensions, time consumption during each phase of the harvesters work, machine movements, and fuel consumption. We
defined the time phases in the process-data model using the
idealized work cycle (Figure 2). The time study material
was recorded from Ponsse, Timberjack, and Valmet harvesters and comprised 200 stems per harvester. The structure of the original process-data model is described in
Figure 3.
The model consists of three hierarchical levels: the level
1 work phases in the hierarchy, the work cycle elements
within these phases (level 2 phases), and the components of
these work cycle elements (level 3 phases). The total work
time for each processed tree equals the combined time
consumption of the level 1 work phases. In levels 2 and 3,
the level 1 work phases are subdivided into smaller work
cycle elements, and the time consumption of each level 1
work phase equals the sum of the work cycle elements at
lower levels of the hierarchy. In the original model, all work
phases are considered to be separate, which means that the
time consumptions do not overlap.
In level 1 of the hierarchy, the work phases are gripping
the stem, felling, and processing. Tables 1 and 2 define
the start and end points of the level 1 work phases and
their work cycle elements. The time consumption during
gripping the stem is calculated as an average value for the
processed trees at each working location or in each stand,
whereas felling and processing times are recorded for
each tree. Here, we used the definition of Kariniemi
(2006), who described the working location as an idealized
Forest Science 59(4) 2013

475

Figure 3. Flowchart for a process-data model describing the relationships between the different work
phases during a harvesters work (Kariniemi and Vartiamaki 2010).
Table 1.

Definition of the time phases used for manual recording.

Work phase

Moving forward
Extend the boom and grip
Felling

Felling and bunching

Processing (delimbing and cross-cutting)
Clearing
Stacking logs
Piling of slash
Bringing the top to the extraction trail
Moving backward
Position the boom forward

Definition
This phase was recorded when the harvester was driving forward, but not when the
harvester was in motion during the felling or processing work phases.
This phase started when the boom began to swing toward a tree and ended when
the chain saw began the felling cut; thus, this phase also includes positioning the
harvester head at the base of the tree.
This phase started when the felling cut began and ended when the feeding and
delimbing of the stem started. Felling included the duration of boom movement
while the head was holding a cut tree to move it to a processing site at a
maximum distance of 3 m from the base of the tree.
This phase included the duration of felling the tree and moving the felled tree to a
processing site located more than 3 m from the base of the tree.
This phase started when the stem began feeding through the harvester head and
ended when the operator lifted the harvester head to an upright position after the
final cross-cut through the stem.
This phase included removal of undergrowth and unmerchantable trees from
around standing trees that must be felled.
This phase included gathering the logs into piles along the extraction trail.
This phase was recorded whenever slash was piled as a separate work phase (i.e.,
not as part of the processing phase).
This phase included bringing unmerchantable tops of stems to the extraction trail
after the final cut to produce the last log.
This phase included the period when the harvester was moving backward, but not
when the harvester was in motion during the felling or processing work phases.
This phase occurred when the operator moved the harvester head to the front of
the machine before moving forward.

area within the reach of the boom, in which a skilled

operator can fell and process all trees without moving the
harvester.
In level 2, the level 1 work phases are subdivided into
476

Forest Science 59(4) 2013

five shorter work cycle elements and four pauses. Tables 1

and 2 provide details of the level 2 time elements. If
positioning occurs while the harvester is moving, the whole
working time is registered as part of the moving phase.

Table 2.

Definition of the time phases used for automatic recording.

Work phase

Gripping the stem

Felling
Sawing during felling
Processing
Driving during processing
Sawing during processing
Using the boom during felling and processing
Simultaneous driving and using the boom during processing

During the gripping the stem phase, when the boom or the
harvester head is motionless due to reasons such as work
planning by the operator or a machine breakdown, the time
is included in the pause 1 phase. Felling is split into felling
and pause 2. Felling starts when the felling cut begins and
ends when the stem feeding starts. Pause 2 is defined as a
time phase during which the machine, boom, and harvester
head are motionless. Processing is divided into stem feed,
pause 3a, arrangement of the products, and pause 3b. The
processing time of each log is split into feeding the stem and
pause 3a. Feeding the stem starts when the stem begins
moving through the harvester head and ends when the
feeding of the log stops. Pause 3a is the time phase during
log processing when the machine, the boom, and the harvester head are motionless. Arrangement of the products
includes operation of the boom and harvester head, and
pause 3b is the time phase without any harvester operations.
Arrangement of the products and pause 3b occur immediately after the processing of each log.
In level 3 of the hierarchy, the moving phase in level 2 is
divided into forward and backward movements. The positioning work phase equals the sum of the extend the boom
phase and the other 1 phase. Extend the boom begins
when the boom starts to swing toward a tree and ends when
the harvester head is resting against a tree. Other 1 includes
phases for clearing undergrowth and for arrangement of the
products around a tree to be felled or repositioning the head
to avoid an obstacle such as a large rock that prevents the
head from reaching the tree. The felling phase includes
the felling cut and felling control phases. Felling cut means
the cut that fells a standing tree, whereas felling control
means moving the felled tree to the position where it will be
processed. The feeding the stem work phase is broken down
into four phases: delimbing, reversing, cross-cutting, and
other 2. During delimbing, the branches are removed by
feeding the stem through the harvester head while the harvester head is moving forwards. Reversing occurs when the
harvester head is moving backward along the stem. Crosscutting includes time consumption during the cross-cut that
produces each log. Other 2 includes work not involving
delimbing or cross-cutting, such as piling logs. The arrangement of the products phase is split into bunching and other
3 phases. Bunching includes moving the stem to the most
convenient position for cross-cutting so that the logs will
form a single pile. Other 3 includes sorting the logs after the

Definition
Starts when the harvester or boom start to move and ends when
the felling cut begins.
Begins when the felling cut starts and ends when the stem
feeding starts.
Duration of the felling cut during felling.
Time from the start of feeding of the first log to ejection of the
final log from the head.
Harvester movement during the processing work phase.
Total time for processing the stem into logs.
Using the boom during the felling and processing phases but
excluding moving.
Using the boom at the same time as processing and moving.

feeding the stem phase and moving tops and branches. The
pause 1, pause 2, pause 3a, and pause 3b phases are split
into shorter phases based on their duration: break for a
pause 3 seconds, rest for a pause longer than 3 seconds
but shorter than 30 seconds, and stoppage for a pause 30
seconds.

Adjustment of the Process-Data Model

In the three tests that provided the data used in our
second and third analyses, we minimized the variability in
the harvesting conditions in the study stands: the terrain was
flat and the tree size, tree species composition, and stem
density (number per unit area) were as similar as possible.
Time study plots in which 45 or 60 minutes of operating
time were monitored were assigned to six harvester operators. Six time study plots from two first-thinning stands
and three time study plots from a clearcutting stand were
selected for each operator. We used data provided by
Vaatainen et al. (2005) for the time for each work phase of
the harvesting system. We observed and timed each phase
of the harvesters work for each processed tree (for a total of
1,776 stems). The effective work time was measured to the
nearest 0.1 second using both the handheld field computer
and the harvesters automatic datalogger.
In the first manual recording test, the observer followed
the steps of the work cycle presented in Figure 4. The
definitions of these work phases are presented in Table 1.
The extend the boom and grip, felling, felling and bunching,
processing, and bringing the top to the extraction trail
phases were defined as the level 1 work phases for each tree.
The moving forward, clearing, stacking logs, piling of slash,
reversing, and positioning the boom forward phases were
complementary work phases that were not analyzed for each
processed tree. In the manual time study, the time for each
of these work phases was recorded separately.
In the second test, automatic recording followed the steps
of the work cycle without including the pause work phases
that are included in the process-data model in Figure 3. The
level 1 work phases were gripping the stem, felling, and
processing. The test concentrated on the processing phase,
and the definitions of the phases are described in Table 2.
Total processing time included driving, sawing during processing, and boom use. Simultaneous driving and using the
Forest Science 59(4) 2013

477

Figure 4.

Flowchart for work phases describing the manually recorded work cycle.

boom during processing was also included in the total

processing time.

Automatic Time Study Methodology

A good research hypothesis and research questions are
integral elements in conducting experimental research because they steer the process of collecting study material,
producing results, making syntheses, and writing the study
report (Dyer and Wilkins 1991). To modify the original
process-data model, we asked the following research
questions:
1.
2.

3.

How does the process-data model describe the automatically recorded components of the work cycle?
Which time consumption components that we recorded manually can be added into the original
process-data model?
What aspects of the process-data model should be
improved based on the answers to research questions
1 and 2?

In this study, we used principal components analysis to

adjust the process-data model. We used this approach to
reduce the variation contained in the measured variables
into principal components that were not correlated with
each other but that explained as much as possible of the
overall variation. Our goal was to find combinations of
work cycle time elements that could be interpreted as dimensions of the overlapping work phases. This analysis
gives every variable a weight that reveals its position within
and its impact on the overall work cycle. The component is
478

Forest Science 59(4) 2013

also given an eigenvalue, which represents the relative

proportion of the overall variation that a component can
explain. We chose solutions that included components with
an eigenvalue greater than 1 (Kaiser 1960) and used the
scree test to determine which factors to retain (Cattell 1966).
We used the Varimax rotation provided by SPSS-X software (SPSS, Inc. 1988) to minimize the number of variables
with high loadings (i.e., high weights) for each factor and
thereby simplify interpretation of the factors.

Results
Automatically Recorded Times
In the first test, we examined the work phases of the
conventional process-data model (Figure 3). Table 3 summarizes the principal components of these work phases in
the improved process-data model. The overall work components in the actual automatic recording can be divided
into gripping the stem, manual processing, and automatic
processing. The gripping the stem component had only one
separate work phase that was not selected for any other
component (gripping the stem), whereas the manual and
automatic processing components both included several
work phases. These work components were not congruent
with the level 1 work phases of the original process-data
model (gripping the stem, felling, and processing). In addition, two of the three level 1 work components included
overlapping work phases.
The manual processing component was split into driving
during processing, using the boom during felling and processing, simultaneous driving and using the boom during

Table 3.

Results of the principal components analysis of the timber-harvesting phases with automatic recording.
Component
Variable

Gripping the stem

Driving during processing
Using the boom during felling and processing
Simultaneous driving and using the boom during processing
Felling
Sawing during felling
Sawing during processing
Processing
Eigenvalue
Proportion of the variation explained (%)

II

0
0.924a
0.568
0.905a
0.661a
0
0
0
2.4
36.2

III

0
0
0.651a
0
0
0.498a
0.813a
0.859a
2.1
21.5

Communalities
a

0.925
0
0
0
0
0.373
0
0
1.1
13.5

0.859
0.874
0.795
0.826
0.528
0.397
0.667
0.753
Total of components IIII
71.2

A Varimax rotation with Kaiser normalization was used in the principal components analysis (weights of 0.3 have been replaced with a weight of 0).
Interpretation of the principal components: I, manual processing; II, automatic processing; III, gripping the stem.
a
The highest weightings for each main component.

processing, and felling. The felling phase within the manual

processing component was equivalent to the felling phase in
the process-data model. The simultaneous driving and using
the boom during processing phase was congruent with the
level 2 phase of log processing in the original model. The
driving during processing phase was an important variable
in this analysis, but it was not defined in the original model.
The automatic processing component was split into processing, using the boom during felling and processing,
sawing during processing, and sawing during felling. The
sawing during processing phase was equivalent to the crosscutting phase in the original model. The using the boom
during felling and processing phase was congruent with the
bunching phase of the original model. However, the using
the boom during felling and processing phase did not fit into
the original model because it included simultaneous machine operations with the manual processing. The sawing
during felling overlapped both the manual processing and
the gripping the stem phases.

they could be added in the improved process-data model.

Table 4 summarizes the principal components of these
work phases. The level 1 work phases in the original
model (gripping the stem, felling, and processing) were
congruent with three of the components revealed in the
manual recording. However, clearing was revealed as an
important additional work component. In level 2 of the
original model, the moving and positioning phases included the same operations observed in the manual recording: position the boom forward, moving forward, and moving backward. At the same level, the felling phase included
the felling phase and the felling and bunching (3 m)
phase. The processing phase in the original model included
one of the same operations revealed by the manual recording (cross-cutting and delimbing). However, the manual
observations included the extend the boom and grip phase
that was included in the positioning phase of the original
model.
In the original model, the extend the boom phase (level
3) and the positioning phase (level 2) diverged from the
position the boom forward phase of the manual recording
(Figures 3 and 4). This is because the positioning phase in
the original model only includes boom movements to fell a

Manually Recorded Time Consumption

In the second test, we examined the manually recorded
components of time consumption to determine whether
Table 4.

Results of the principal components analysis of the timber harvesting with manual recording.
Component
Variable

Moving forward
Extend the boom and grip
Felling
Cross-cutting and delimbing
Clearing
Bringing the top to the strip road
Moving backward
Position the boom forward
Felling and bunching (3 m)
Clearing and positioning
Clearing and felling
Eigenvalue
Proportion of the variation explained (%)

I
0
0
0
0
0.0765a
0
0
0
0
0.862a
0.708a
1.8
17.1

II
0
0
0.921a
0
0
0
0
0
0.910a
0
0
1.7
15.6

III
a

0.788
0
0
0
0
0
0.612a
0.752a
0
0
0
1.6
14.7

IV

Communalities

0
0.781
0
0.723a
0
0.205
0
0
0
0
0
1.2
11.5

0.664
0.615
0.866
0.539
0.628
0.058
0.394
0.589
0.862
0.766
0.505
Total of components IIV
58.9

A Varimax rotation with Kaiser normalization was used in the principal components analysis (weights of 0.2 have been replaced with a weight of 0).
Interpretation of the principal components: I, clearing; II, felling; III, gripping the stem; IV, processing.
a
The highest weightings for each main component.
Forest Science 59(4) 2013

479

tree, whereas in the manual recording, the position the boom

forward phase was recorded separately when the operator
steered the harvester head to the front of the machine before
moving to the next working location. The principal component analyses included this phase in the gripping the stem
component (Table 4).

includes boom movements to fell a tree. In our improved

model, it was included in the gripping the stem component.
The bringing the top of the stem to the extraction trail phase
could be included under the processing phase (level 1) of
the original model. In our improved model, this work phase
was under the arrangement of the products work component. Furthermore, the extend the boom and grip the stem
(M) work phase was included in the positioning work
component.
The felling component included the felling (M) work
phase and the felling and bunching (M) work phase. On the
other hand, the felling (automatically recorded [A]) phase
was included in the manual processing component. The
sawing during felling phase was included in the felling cut
phase of the original model, but in the improved model, it
was included in the clearing component. These results indicate different timing allocation between the manual and
automatic recordings. In the original model, the manually
recorded clearing phase was included in the other 1 phase.
This was possible because in the model, the total gripping
the stem time is usually calculated as the average value for
the stems at each working location or for the whole stand. In
our model, clearing (M), clearing and positioning (M), and
clearing and felling (M) were included in the clearing component. These results indicate a different hierarchical structure between the original and improved models. The hierarchical structure of the improved model did not include
complementary work phases.
The processing phase (level 1) in the original model
included work phases from the start of feeding of the first
log to ejection of the final log from the head. Therefore, the

Process-Data Model
In the third test, we investigated the changes required to
improve the original process-data model based on the answers to research questions 1 and 2. The potential work
phases that could improve the model were identified by
means of principal components analysis (Table 5), which
allowed us to combine the important work phases from the
manual and automatic recordings. The main work components were automatic processing, manual processing, clearing, moving, gripping the stem, felling, positioning, and
arrangement of the products. The level 1 phases in the
original model (gripping the stem, felling, and processing)
were congruent with the main work components revealed by
the principal components analysis. In addition, the manual
observations revealed manual processing, clearing, moving,
positioning, and arrangement of the products as additional
level 1 work phases. In level 2 of the original model, the
moving and positioning work phases included the extend
the boom and grip (manually recorded [M]), moving forward (M), and moving backward (M) phases.
The position the boom forward (M) phase occurs before
the harvester starts to move to the next working location.
This phase could not be incorporated in the original model
because the definition of positioning in the model only
Table 5.

Results of the principal components analysis of the timber harvesting with both manual and automatic recording.
Component
Variables

Moving forward, M
Extend the boom and grip, M
Felling, M
Cross-cutting and delimbing, M
Clearing, M
Bringing the top to the extraction trail, M
Moving backward, M
Position the boom forward, M
Felling and bunching (3 m), M
Clearing and positioning, M
Clearing and felling, M
Gripping the stem, A
Driving during processing, A
Using the boom during felling and
processing, A
Simultaneous driving and using the boom
during processing, A
Felling, A
Sawing during felling, A
Sawing during processing, A
Processing, A
Eigenvalue
Proportion of the variation explained (%)

II

III

0
0
0
0
0
0
0
0
0
0.917a
0
0
0
0
0.714a
0
0
0
0
0
0
0
0
0
0
0.415
0
0
0
0.792a
0
0
0.750a
0
0
0
0
0.897a
0
0.575a 0.504 0.495
0

0.896a

0
0.639a
0
0
0
0.475a
0.795a
0
0
0.910a
0
0
2.8
2.5
2.3
14.7
13.3
12.1

IV

VI

VII

0.793a
0
0
0
0
0
0.814a
0
0
0.950a
0
0
0
0
0
0
0
0
0
0
0
0
0
0.977a
0.540a
0
0
0
0.649a
0
0
0
0
0.834a
0
0
0
0
0
0
0
0
0
0
0.762a
0
0.397
0
0
0
0
0
0
0
0
0

Communalities
0.690
0.699
0.942
0.849
0.591
0.960
0.350
0.590
0.928
0.682
0.587
0.812
0.873
0.855

0.834

0
0
0
0
2.1
10.9

0
0
0
0
1.7
8.8

0.502
0.313
0
0
1.5
7.7

0
0
0
0
1.1
5.3

0.685
0.421
0.636
0.843
Total of components IVII
72.8

A Varimax rotation with Kaiser normalization was used in the principal components analysis (weights of 0.3 have been replaced with a weight of 0).
The highest weightings are presented in boldface for each main component. Interpretation of the principal components: I, automatic processing; II, manual
processing; III, clearing; IV, gripping the stem; V, felling; VI, positioning; VII, arrangement of the products.
a
The highest weightings for each main component.

480

Forest Science 59(4) 2013

driving during processing (A) work phase could be added

into either the other 2 or other 3 tertiary phases. Furthermore, the sawing during processing (A) phase could be
added in the cross-cutting tertiary phase. In the improved
model, the automatic and manual processing components
systematically replace the processing level 1 phase. Therefore, the driving during processing (A) phase was included
in the manual processing component and the sawing during
processing (A) phase was included in the automatic processing component. Furthermore, the using the boom during
felling and processing (A) phase did not fit into the original
model because it included significant overlapping durations
due to simultaneous machine operations but could be included in the improved model.

Discussion
Assessment of the Study
We recorded the work phases of a single-grip harvester
in parallel using automatic and manual recording techniques. This let us compare the information value of both
techniques. Statistical methods were successfully used in
analysis, although the experiment of this study does not
offer the possibility for statistical generalization. The experimental study strategy applied in our study was to describe the potential method for automatic and manual timing
to reach a better understanding of the automatic time study
method. Although we collected enough data to have confidence that our results are statistically valid (i.e., that they
can be generalized to other machines and stands), the purpose of our study was not to collect statistics on cycle times,
but rather to identify the optimal allocation of cycle times
within an improved process-data model. Using the strategy
of Dyer and Wilkins (1991), we were able to analyze the
automatic and manual timing data from the studies by
Vaatainen et al. (2005) and Kariniemi and Vartiamaki
(2010).
The problem of discrete work phases in the manual
timing was avoided by analyzing additional subphases of
the level 1 work phases using data collected automatically
by a datalogger attached to the machines data bus. There
was significant overlap among the work phases. We therefore used principal components analysis to provide a statistical basis for the use of partial scales (i.e., partially overlapping work phases). This analysis revealed several key
underlying factors that would not have been identified using
the original process-data model or by using completely
automated time study methods such as the one introduced
by McDonald and Fulton (2005).

Improved Process-Data Model

For cut-to-length harvesting systems, Table 5 presents an
improved version of the process-data model presented in
Figure 3. The new work phase classification was independent of the manual timing techniques, because the time per
work phase was not recorded separately and because certain
secondary work phases were omitted in the manual recording. For example, the automatic time study method recorded
the duration of using the boom during felling and process-

ing. This overlapping operation could not be included in the

original model because the work phases in that version of
the model are constructed separately for use in modeling
studies (Olsen et al. 1998, Spinelli et al. 2010) aimed to
determine the general relationships between time consumption and parameters of the working conditions; such
studies require that the work cycle in the time study be
constructed from regularly repeating elements without overlapping subphases. Our improved process-data model is
based on a more systematic and clear hierarchical structure
of the work phases, which can account for both separate and
overlapping work phases (Table 5). There will always be
differences between data-collection techniques, making it
challenging to produce time consumption information in a
consistent format.
One advantage of the improved process-data model was
that it enabled combinations of the information obtained
using automatic and manual recording. An additional benefit of our analysis is that it includes an analysis of delays,
which are one of the major factors that limit productivity
and are, therefore, an integral part of most time studies (e.g.,
Spinelli and Visser 2008). The division of work cycle time
elements in the improved model allows pause times to be
recorded by a datalogger and included in the model as new
work phases. Furthermore, automatically recorded pause
times can be incorporated in the hierarchy of the improved
model.
As Table 5 shows, the duration of the using the boom
during processing and felling phase indicated the existence
of overlapping and simultaneous work phases. This confirms the results of Vaatainen et al. (2005), Kariniemi
(2006), and Ovaskainen (2009). In those studies, the overlapping durations of simultaneous work phases were important indicators for explaining operator performance and
motor sensory abilities. The simultaneous work phases
could also be used to identify the human factors that influence the performance of a human-machine system (Palander
et al. 2012). Therefore, the proportion of simultaneous work
phases should be more carefully accounted for in future
research. However, measuring simultaneous phases with a
handheld field computer is a challenging task because the
simultaneity of different operations requires the presence of
two or more observers (Nuutinen et al. 2011). In this respect, our improved model allows a highly detailed work
cycle projection and increases our ability to understand the
structure of the human-machine system. The findings from
such a study could be used in operator training to make the
overall system more efficient.
The rapid evolution of information technology can allow
managers to generate local adaptations of process-data models in individual stands. Our adjusted model provides both a
conceptual-theoretical basis and a practical basis to reconcile the results of time studies based on different methods.
Therefore, the adoption of our harmonized time study protocol would prevent misunderstandings. Adoption of this
approach would also facilitate the production of more internationally comparable work study reports. These suggestions are in accordance with recent methodological experience gained from the adaptive control of a human-machine
system (Palander et al. 2012). For advanced work study
Forest Science 59(4) 2013

481

techniques, it is necessary to adjust the process-data model

to local work conditions using an automatic time study
method. The advantage of the improved model is that it can
be adapted to human-machine systems, depending on the
study subject or measurement technique. For example, for
short monitoring or control periods, the level 1 work phases
could be broken into segments of the effective work time,
excluding pause times.

Advantages of the Automatic Time Study

Method
In our study, the automatic time study method was developed by analyzing the work phases under similar working conditions. It must be noted that, to date, the presence of
a researcher has usually been required to detect unexpected
situations during harvesting. The use of automatic recording
solves this problem and gives the researcher extra time to
investigate matters such as conditions at the logging site that
explain overall productivity and the operators working
technique. For example, Vaatainen et al. (2005) included
boom movements and moving between working locations in
the gripping the stem phase recorded by the PlusCan datalogger. These components could be used to describe the
effect of the working conditions. For example, the gripping
the stem time increases when the terrain is difficult to
traverse or when tree density is low.
The phases of gripping the stem and felling that are
defined in level 3 of the model can be recorded manually
using a handheld field computer, except for work phases
with short durations (Figure 3). Nuutinen et al. (2008) found
that an observer cannot reliably measure work phases
shorter than 3 seconds. This is also true for the log-level
operations during processing in level 3 of the models
hierarchy, which cannot be recorded manually, especially
when large stems and a large number of stems are measured.
Such observations are limited by observer fatigue during the
course of a long time study (Nuutinen et al. 2008). Our
automatic recording let us analyze short work phases, such
as the sawing during processing phase. This possibility was
also used by Nuutinen et al. (2010) to compare the efficiency of different feed rollers in a harvester head.
We found differences in the durations of the work phases
between the manually and automatically collected data.
These results indicated that the automatic time study
method provides more systematic and accurate recording of
the work phases. These differences should be taken into
account when the advantages of different data collection
methods are compared. Nuutinen et al. (2008) found that an
observers skill and experience affected measurement accuracy in manual time studies and thereby affected the results,
especially during intensive time studies of harvester operations. Therefore, the accuracy of manual timing is limited
when such machines are monitored. In addition, manual
recording cannot produce sufficiently detailed and diverse
information for the log-level durations of work phases.
There is no doubt that automatic recording enables the
collection of larger volumes of data at lower cost than with
traditional manual observations. Palander et al. (2012) demonstrated this by automatically recording more than 50 work
482

Forest Science 59(4) 2013

study variables and using computerized data mining to

select the most important work conditions and work phases.
Nuutinen et al. (2010) also performed a highly detailed and
accurate projection of fuel consumption and processing
times for 7,400 stems using a harvesters automated datalogger. Our approach worked well because the time study
method allowed efficient adjustment of the original processdata model. As recent advanced studies have suggested
(Nuutinen et al. 2010, Palander et al. 2012), the entire
data-mining phase, including the transfer of data for further
analysis, can be automated using automatic dataloggers
combined with methods such as the one developed in the
present study. The insights provided by our study suggest
that the methodology represents a powerful core for a data
management system and that it has great potential to significantly improve the efficiency of time studies of humanmachine systems.

Conclusions
In this article, we presented the results of three representative time studies of single-grip harvesters. We used the
results to develop an automatic time study method with an
improved ability to capture the key components of the
harvesters work cycle. This method can be used to adjust
the original process-data model of Kariniemi and Vartiamaki (2010) to account for different time studies under
similar harvesting conditions. To adjust the model for different stands, managers can reorganize the accurate time
data gathered by this systematic method. The benefit of our
approach is that it can identify the most important work
phases from large amounts of time study data. The principal
components analysis identified the optimal reorganization
of the model using an objective and statistically valid
method. The improved process-data model is superior to the
original because it can record simultaneous work phases
that overlap to varying degrees. This is an important advantage for mechanized and semiautomated work operations,
for which both the machine operator and the harvesters
computer may control certain actions. Adjustment of the
model to improve data recording accuracy has great potential in future forestry work, but this must be confirmed
through additional time studies under different working
conditions.

Literature Cited
ARLINGER, J., S. BERGEK, L. HULT, B. LARSSON-SNYGG, B, MORENIUS, AND J. SONDELL. 2010. Operational monitoring of forest
machines under working conditionDefinitions and implementations to time concepts and repair causes with in Stanford
frameworkVersion 2.9. Skogforsk, Uppsala, Sweden. 19 p.

ANEN

ARLINGER, J., J.J. MOLLER

, T. RAS
, AND J.A. SORSA. 2008.
StanForD 2010. Benefits, needs and requirements for users and
manufacturers. Metsateho Report 201. Available online at
www.metsateho.fi/files/metsateho/Raportti/Raportti_201_TR_
Report_StanForD_tr_ym.pdf; last accessed May 28, 2012.

BJORHEDEN
, R. 1991. Basic time concept for international comparisons of time study reports. J. For. Eng. 2(2):3339.
CAN IN AUTOMATION. CAN history. Available online at
www.can-cia.de/index.php?id161/; last accessed May 28,
2012.

CATTELL, R.B. 1966. The scree test for the number of factors.
Multivar. Behav. Res. 1:245276.
DYER, W.G., AND A.L. WILKINS. 1991. Better stories, not better
constructs, to generate better theory: A rejoinder to Eisenhardt.
Acad. Manage. Rev. 16(3):613 619.
EISENHARDT, K.M. 1989. Building theories from case study research. Acad. Manage. Rev. 14(4):532550.
EISENHARDT, K.M. 1991. Better stories and better constructs: The
case for rigor and comparative logic. Acad. Manage. Rev.
16(3):620 627.
ELIASSON, L. 1998. Analyses of single-grip harvester productivity.
PhD thesis, Swedish University of Agricultural Sciences,
Ume, Sweden. 24 p.
FPINNOVATIONS. 2012. MultiDAT general description. Available
online at tinyurl.com/26u7585; last accessed May 28, 2012.
HARSTELA, P. 1988. Principle of comparative time studies in
mechanized forest work. Scand. J. For. Res. 3:253257.
INTERNATIONAL LABOUR OFFICE. 1981. Introduction to work
study. International Labour Office, Geneva, Switzerland. 442 p.
INTERNATIONAL UNION OF FORESTRY RESEARCH ORGANISATIONS.
1995. Forest work study nomenclature. International Union of
Forestry Research Organisations, Vienna, Austria. 16 p.
JOHN DEERE. 2008. TimberOffice: Timberlink. Available online at
www.timberoffice.com/english/products/timberlink/; last accessed May 28, 2012.
KAISER, J.L. 1960. The application of electronic computers to
factor analysis. Educ. Psychol. Measure 20:141151.
A , K., E. RONKK

, AND S.I. GUMSE. 2004. Productivity and

KARH
O
cutting costs of thinning harvesters. Int. J. For. Eng.
15(2):4356.
KARINIEMI, A. 2006. Operator-specific model for mechanical harvestingCognitive approach to work performance. PhD thesis,
University of Helsinki, Helsinki, Finland. 126 p (in Finnish,
with English abstract).
. 2010. Hakkuukoneen tietoKARINIEMI, A., AND T. VARTIAMAKI
jarjestelma tutkimustiedon lahteena [The digital data of singlegrip harvesters as a work study material]. Metsateho. Metsatehon raportti 212. Available online at www.metsateho.fi/
files/metsateho/Raportti/Metsatehon_raportti_212_Hakkuukoneen
_tietojarjestelma_tutkimustiedon_aka.pdf; last accessed May 28,
2012.
KELLOGG, L.D., AND P. BETTINGER. 1994. Thinning productivity
and cost for mechanized cut-to-length system in the Northwest
Pacific coast region of the USA. Int. J. For. Eng. 5(2):4352.
MCDONALD, T.P., AND J.P. FULTON. 2005. Automated time study
of skidders using global positioning system data. Comput.
Electron. Agr. 48:19 37.
NAKAGAWA, M., J. HAMATSU, T. SAITOU, AND H. ISHIDA. 2007.
Effect of tree size on productivity and time required for work
phases in selective thinning by a harvester. Int. J. For. Eng.
18(2):24 28.
NORDIC FOREST STUDY COUNCIL. 1978. Forest work study nomenclature. Nordic Forest Study Council. 130 p.
NURMINEN, T., H. KORPUNEN, AND J. UUSITALO. 2006. Time
consumption analysis of the mechanized cut-to-length harvesting system. Silva Fenn. 40(2):335363.
A , J. LAITILA, P. JYLHA , AND S. KESKINEN.
NUUTINEN, Y., K. KARH
2011. Productivity of whole-tree bundler in energy wood and
pulpwood harvesting from early thinnings. Scand. J. For. Res.

26(4):329 338.
AINEN

NUUTINEN, Y., K. VA AT
, A. ASIKAINEN, R. PRINZ, AND
J. HEINONEN. 2010. Operational efficiency and damage to sawlogs by feed rollers of the harvester head. Silva Fenn.
44(1):121139.
AINEN

, J. HEINONEN, A. ASIKAINEN, AND

NUUTINEN, Y., K. VA AT

D. ROSER
. 2008. The accuracy of manually recorded time study
data for harvester operation shown via simulator screen. Silva
Fenn. 42(1):6372.
OLSEN, E., M. HOSSAIN, AND M. MILLER. 1998. Statistical comparison of methods used in harvesting work studies. Oregon
State University, Research Contribution 23, Forest Research
Laboratory, Corvallis, OR. 31 p.
OVASKAINEN, H. 2009. Timber harvester operators working technique in first thinning and the importance of cognitive abilities
on work productivity. PhD thesis, University of Eastern Finland, Joensuu, Finland. 62 p.
OVASKAINEN, H., J. UUSITALO, AND T. SASSI. 2006. Effect of edge
trees on harvester positioning in thinning. For. Sci. 52(6):
659 669.
PALANDER, T., H. OVASKAINEN, AND L. TIKKANEN. 2012. An
adaptive work study method for identifying the human factors
that influence the performance of a human-machine system.
For. Sci. 58(4):377389.
PALMROTH, L. 2011. Performance monitoring and operator assistance systems in mobile machines. PhD thesis, Tampere University of Technology, Tampere, Finland. 114 p.
PELTOLA, A. 2003. IT-time for Mechanised Forest Work Study.
2nd Forest Engineering Conference 1215 May 2003, Vaxjo,
Sweden. Skogsforsk Arbetsrap. 536:107112.
PUTTOCK, D., R. SPINELLI, AND B. HARTSOUGH. 2005. Operational
trials of cut-to-length harvesting of poplar in a mixed wood
stand. Int. J. For. Res. 16(1):39 49.
SAMSET, I. 1990. Some observations on time and performance
studies in forestry. Commun. Norwegian For. Res. Inst. 45.5.
80 p (in Norwegian with an English summary).
SKOGFORSK. 2010. StanForD. Available online at www.skogforsk.
se/en/About-skogforsk/Collaboration-groups/StanForD/; last
accessed May 28, 2012.
SPINELLI, R., B.R. HARTSOUGH, AND N. MAGAGNOTTI. 2010. Productivity standards for harvesters and processors in Italy (statistical data). For. Prod. J. 51(4):54 61.
SPINELLI, R., AND R. VISSER. 2008. Analyzing and estimating
delays in harvester operations. Int. J For. Eng. 19(1):36 41.
SPSS, INC. 1988. SPSS-X users guide, 3rd ed. SPSS Inc., Chicago,
IL.
TIKKANEN, L., H. OVASKAINEN, T. PALANDER, AND L. VESA. 2008.
TimberLink as a tool for measuring the fuel consumption of a
harvester. P. 70 71 in The Nordic-Baltic conference on forest
operations, Suadicani, K. (ed.) Forest & Landscape, Working
Papers 30, Denmark. Available online at www.docstoc.com/
docs/95433278/The-Nordic-Baltic-Conference-on-ForestOperations; last accessed May 28, 2012.
AINEN

, K., H. OVASKAINEN, P. RANTA, AND A. ALA-FOSSI.

VA AT
2005. The significance of harvester operators tacit knowledge
on cutting with a single-grip harvester. Communications of the
Finnish Forest Research Institute 937. 100 p (in Finnish with an
English summary).

Forest Science 59(4) 2013

483

Reproduced with permission of the copyright owner. Further reproduction prohibited without
permission.