Time Study Standards First Edition
Standards for Time Studies for the South African Forest Industry
First Edition 5 February 2014 Project team: Pierre Ackerman, Elizabeth (Lise) Gleasure, Simon Ackerman, Brad Shuttleworth
Summary
This South African Standard for Timestudies will provide a common and standard timestudy methodology for the South African Forest Industry; a protocol that does not currently exist. Its implementation will serve the purpose of aligning the South African Forest Industry with international forest operations development and assist with the “modernisation” of the Industry’s forest operations. The concept of modernisation essentially includes updating forest operations in terms of both mechanisation and other modern systems improvements with the goal of improving wood/fibre yield, wood/fibre quality and reducing production costs to remain locally and internationally competitive. The Standard has been compiled by those with specific expertise in work and timestudies, particularly the statistical analysis component and machine costing. The Standard, with the inclusion of an internationally standardised Machine Costing Model, was developed based on accepted and validated international Timestudy standards, protocols and literature. This Protocol is envisaged to be a state of the art model to benefit the South African Forest wood supply chain. The Standard will be webbased and will guide the user stepbystep through the set up and execution of time studies and their application in Operations Research analysis. The standard deals with the setting of timestudy objectives to ensure that time and resources are used efficiently and help to develop the desired results. Three types of studies, observational, experimental and modelling, are introduced. Different techniques are provided to control bias (i.e., systematic error) including randomisation and blocking. The Standard contains sections on experimental study design, data collecting methodologies including sample size calculations; time study models; selecting an appropriate time study technique; statistical analysis and methods to best analyse the data collected; and ways to use and proceed with the results achieved through linkage with a machine costing model. The user will also be able to calculate machine availability, utilisation and systems efficiency ratios that are useful in determining systems efficiency. Background data forms, a terrain classification, templates to create data collection forms for the user’s study and a brief discussion on available time study software and equipment are also included. Included in the Standard is a Timeconcepts model developed by the International Union of Forest Research Organisations (IUFRO), useful for the precise division of common time elements included in all work and production systems. The Standard also describes in detail the six different scopes of time studies, ranging from wide to narrow. These studies are shiftlevel, plot level, cycle level, time and production count, working sampling and the element level. Each study has different strengths and weaknesses and requires a specific technique which is discussed. A statistical analysis manual is also in the drafting stages and will aid the user through conducting their analysis and interpreting the results.
Table of Contents
1.0 Introduction
1.1 Background 1.2 From Work Study to Time Study
2.0 Setting up a Timestudy
2.1 Developing a Study Goal and Objective 2.2 Study Classification and Experimental Design 3 2.2.1 What type of study do you need?/span> 3 2.2.2 Observational study 2.3 Experimental Study Designs 2.3.1 Monofactorial Random Design 2.3.2 Multifactorial Random Design 2.3.3 Monofactorial Block Design 2.3.4 Multifactorial Block Design 2.3.5 Monofactorial Latin Square Design 2.3.6 Multifactorial Latin Square Design 2.3.7 Splitplot Designs 2.4 Modelling Studies 2.5 Sample size calculations 2.5.1 Pilot Studies 2.5.2 Approximating Sample Size
3.0 Time Models
3.1 Ratio Calculations 3.1.1 Mechanical Availability 3.1.2 Machine Utilisation 3.1.3 Capacity Utilisation 3.1.4 Visualisation of Time Concepts
4.0 Time Study Techniques and Methodologies
4.1 Shift Level Study 4.1.1 Data acquisition methods 4.1.2 Advantages and Drawbacks 4.2 Plot Level Study 4.2.1 Data acquisition methods 4.2.2 Advantages and Drawbacks 4.3 Cycle Level Study 4.3.1 Data acquisition methods 4.3.2 Advantages and Drawbacks 4.4 Time and Production Count 4.4.1 Data acquisition 4.4.2 Advantages and Drawbacks 4.5 Element Study 4.5.1 Data acquisition 4.5.2 Advantages and Drawbacks 4.6 Work Sampling (Instantaneous Observation, and/or Activity Sampling) 4.6.1 Data acquisition 4.6.2 Advantages and Drawbacks
5.0 Machine Element Standardisation
5.1 Standardised Element Lists by Machine: Chainsaw: Harvester: Fellerbuncher: Skidder/agricultural tractor with winch or drawbar (aframe or other): Skidder (grapple): Loader: Loader (either tracked or wheeled): Processor: 31 Truck (timber transport): Yarder: Mulchers and Destumpers: 5.2. Userdefined elements
6.0 Statistical Analysis
7.0 References
1.0 Introduction
The purpose of this protocol framework is to provide a standardised time study methodology for the South African forest industry. This manual has been developed to work in conjunction with the partner computer program to assist in work study development. This program is developed specifically as an extension of this manual and as a way to assist the user through navigating the concepts in the manual relevant to their study objective. This manual will cover setting up a time study, selection of experimental design (2.0), time models and time concepts (3.0), time study methods (4.0), standardised, machinespecific time elements (5.0) and statistical analysis (6.0 – still to be completed). Tools included with this manual are the abovementioned software, study forms – both generic and machinespecific, and a further reading list. The outputs of time study analysis can then be inputted into a costing model developed by the European Union Cost Action 0902. This costing model has been developed by experts from around the globe, including South Africa, and provides an easy to use and internationally regarded way to cost forest operations (Figure 1). The model makes use of internationally accepted and current costing protocols and has been validated by an expert panel. The costing model can be found on the cost website at this link: http://www.forestenergy.org/pages/costingmodel—machinecostcalculation/?PHPSESSID=68b81c040f0688cadc1a350adda16c9c.
Figure 1: Screenshot of costing model developed by the European Union Cost Action 0902. A corresponding manual has been written to support the Microsoft Excel based model.
1.1 Background
Improving operations efficiency is an ongoing need for all industries, including forestry. The South African forest industry faces unique challenges and addressing efficiency in this context is complex. A key tool to address the challenges of efficiency and productivity improvement comes from the discipline of work science; to study work and productivity. Work science is the study of work and its associated measurement including human elements, the machines and other equipment used for work, the organisation of work and the methods of work (Björheden and Thompson, 1995). Work science has a long history with forestry, having developed into an independent field as early as 1920. The origin of work science is often attributed to F.W. Taylor’s (1895) paper titled “A piecerate system being a partial solution of the labour problem” published in the Transactions of the American Society of Mechanical Engineers (Barnes, 1963). Taylor’s emphasis on determining a standard amount of time for a task under certain conditions of measurement forms the basis for improving efficiency (Barnes, 1963). It is from this basis that work study methodologies developed. Work study is the systematic examination of the methods of carrying on activities so as to improve the effective use of resources and to set up standards of performance for the activities being carried out (Kanawaty, 1992). The aim of work study is to examine how activities are carried out to complete a task and the use this information to simplify or modify and then use the activity to reduce unnecessary or excess work (Kanawaty 1992).
1.2 From Work Study to Time Study
A work study is typically broken up into two parts, the method study and then the work measurement (Kanawaty, 1992). A method study is normally the first step in order to determine what the optimal method for completing a task is. A method study is defined as a study where the task is systematically recorded and critically examined to find ways to make improvements to the task completion (Kanawaty, 1992). An example of a change in method may be using three chokermen rather than two with the increased productivity making up for the increased cost of wages. Once a method has been established, then the work measurement can begin. Most commonly, the time study is used to determine the standard time it should take to complete the task using the optimised method. Different time study techniques and scales of study exist; these are detailed in Section 4. The outcome of the time study is typically a measure of productivity per productive machine hour (i.e. 30 m^{3} hour^{1}). This output data is incredibly useful allowing for the creation of machine or operation standards, accurate inputs to preexisting costing models, and potentially the creation of models to predict a machine or operations productivity given certain inputs.
2.0 Setting up a Timestudy
2.1 Developing a Study Goal and Objective
Before any study is undertaken, the objective needs to be determined. The development of a clear objective ensures that time and resources are used efficiently and help to develop the desired results. Examples of work study objectives are:
 Locate inefficiencies in a particular harvesting system
 Determine the productivity of a new operator
 Compare two harvesting systems’ productivity
 Assess a machine’s downtime and find reasons for downtime
 Develop a production model for a specific machine
Once an objective is set, a study can then be designed to achieve this objective.
2.2 Study Classification and Experimental Design
A sound experimental design makes it easier to achieve the objectives of the experiment, as has already been mentioned. The early establishment of experimental design makes it far easier to conduct an experiment, collect the required data, and conduct the statistical analysis required. Although an experimental design can be constructed to achieve most study objectives, for the purposes of simplification, this manual will be divided into three study types: observational studies, experimental studies and modelling studies. Modelling studies in particular can be considered a subset of experimental studies and it can be argued that a modelling study can be obtained from both the observational and experimental studies and therefore is not an independent study type. However, the proposed division of study types allows for greater focus for the user of this manual to be spent on designing for the study’s objective. This section is a guideline on how to design an experiment in the context of an operation time study.
2.2.1 What type of study do you need?
There are three types of studies that can be used, although they are not mutually exclusive. The first is the observational study. In an observational study, variables are not controlled (Magagnotti and Spinelli, 2010; Kanawaty, 1992). This study type serves to describe the current state of a machine, operation or system. The second type of study is the experimental study. This type involves greater control of variables and produces results that are more statistically rigorous. The final classification is a modelling study. This type of study is done to create a model for a given machine, operation or system. Modelling differs mainly in the purpose of using the empirical information for modelling and later simulation (computer implemented modelling). However to keep things simple and for purposes of study, classification in this guideline will treat the three study types as separate entities. Standard units of measure include (see Section 5.1 for standardised elements by machine with units described and defined): m^{3}, tons, tons/m^{3} pmh^{1}/smh^{1}/amh^{1 }
2.2.2 Observational study
An observational study (not to be confused with activity sampling techniques and which are discussed in Section 4.6), also called descriptive study, is typically done to learn more about a specific machine, operator or system. This is the simplest study design as it does not require comparisons with other machines, operators or systems and where variables around the machine or systems function are not controlled. In essence an observational study draws inferences about the possible effect of a treatment on subjects, where the assignment of subjects into a treated group versus a control group is outside the control of the investigator. This is in contrast with experiments, such as randomised controlled trials, where each subject is randomly assigned to a treated group or a control group. The treatment unit is the desired machine, operator or system. Different measurement methods can be used depending on the study’s final objective. Example of Study Objective Determine the productivity of a fellerbuncher. What statistical analysis can be done? Basic calculations include productivity and costs and are calculated using standard units (see Section 5.1) for the given machine, operator or system. Basic descriptive statistics (e.g. means, medians, minimum and maximum values and standard deviations). The confidence intervals can also be determined. What are the strengths and weaknesses of this design type?
2.3 Experimental Study Designs
Experimental designs compare different variables in order to determine differences or establish cause and effect. Because more control of variation (such as slope, machine type, etc.) is required, these designs are usually more complex. Different techniques are used to control bias (i.e. systematic error) including randomisation and blocking. Bias refers to a tendency to over represent or under represent certain parts of the population (Ott, 1993). A factor (treatment) or factors are applied to see the effects. Determining the effect of one factor is referred to as the “main effect” of the factor. For example, if one wanted to see how skidder type, cable or grapple, influences productivity, the main effect examined would be skidder type. When multiple factors are involved, the interaction between the factors may also become significant. For example, if one wanted to see how skidder type (cable or grapple) and skidder engine capacity (e.g. 130kW) influence productivity in combination, first the interaction between skidder type and engine size would be tested, as the hypothesis tested is that there is no factor interaction effects. If the interaction is not significant, the hypothesis can be rejected and therefore it is sufficient to test the main effects (Milton and Arnold, 1999). To put it in simpler language, if the two (or more factors) do not interact statistically then the factors only need to be examined individually. Another key concept to mention here is that of variance. Variance, in layman’s terms, describes the spread of data around the mean (aka the average). For example, Table 1 below shows two different sets of values. Both have the same mean of 2.75; however, the spread of values in set B is much wider than set A; therefore, set B has the larger variance of the two sets. Greater explanation of variance can be found in Section 6.0: Statistical Analysis. Table 1: An example of variance; two sample sets can have the same mean but different variances. Experimental designs are described below and each is discussed in terms of factors and the bias control technique used as well as the strengths and weaknesses of each design. Three basic assumptions need to be adhered to in the analysis with standard linear statistics (i.e. ttest, ANOVA and ordinary linear regression): homoscedacity (statistically similar variances) and independence of data. Should these basic assumptions not be met, advanced statistical analysis is required. It is recommended that the user seek the guidance of a statistical professional in this case. Care should also be taken to either use one operator across all treatments or use similar operators. A confounding factor can quite quickly develop if this factor is ignored. Confounding means that it becomes impossible to find out whether the relationship (or lackthereof) is a result of the block or the treatments themselves) (Clewer and Scarisbrick, 2001). Unless determining whether an operator is more effective than another operator, always ensure any differences between operators are minimal. This section draws on the work of Pretzsch (2009) as well as Clewer and Scarisbrick (2001).
2.3.1 Monofactorial Random Design
A monofactorial random design involves testing (or comparing) one specific factor (Pretzsch, 2009). Bias is controlled through randomisation. This study is conducted to compare one factor under the condition that the study site conditions are homogenous (i.e. they do not vary drastically from each other). Example of Design: As an example, a study could be designed to compare productivity between a grapple and cable skidder. Operators have both been working for the same amount of time, have the same amount of training and can be considered similar. Alternatively, one can study the same operator on both machines to reduce the potential of differences between operators. Site and stand conditions are selected in a way that they do not differ for the two systems. The treatment is therefore skidder type (Cable vs Grapple). What statistical analysis can be done? Basic calculations include productivity and costs and are calculated using standard units for the given machine, operator or system. Basic descriptive statistics and confidence intervals can also be determined. Treatment effects are tested using an Analysis of Variance (ANOVA). What are the strengths and weaknesses of this design type?
2.3.2 Multifactorial Random Design
A multifactorial random design involves testing two or more factors (Pretzsch, 2009). Bias is controlled through randomisation. This study is conducted to compare multiple factors and the study site conditions are homogenous (they do not vary drastically from each other) (Pretzsch, 2009). Example of Design One can design a study to examine the productivity of a cable skidder and a grapple skidder as well as how productivity varies between morning and afternoon shifts.
Cable Morning Shift  Grapple Afternoon shift 
Grapple Morning Shift  Cable Afternoon shift 
What statistical analysis can be done? Basic calculations include productivity and costs and are calculated using standard units for the given machine, operator or system. Basic descriptive statistics and confidence intervals can also be determined. Treatment interactions as well as individual treatment effects are tested using factorial ANOVAs. What are the strengths and weaknesses of this design type?
Strengths  Weaknesses 


2.3.3 Monofactorial Block Design
Monofactorial block design involves testing (or comparing) one specific factor (Pretzsch, 2009). Block designs are used to reduce known systematic variation, (e.g. known changes in slope category, different shift times, etc.). This is done through the technique of blocking where treatments are grouped across the different categories (Cluwer and Scarisbrick, 2001). Systematic bias is therefore controlled through a combination of the development of blocks and the remaining bias is controlled through randomly placing treatments in blocks (Cluwer and Scarisbrick, 2001). In other words it ensures that random effects are avoided that lead to a clustering of repetitions of the same treatment, which would lead to a bias in case of spatial correlations within the experimental site. Example of Design A study is designed to compare the productivity of three operators (the treatment factor is therefore operator), Abe, Bob and Carl. The sites vary depending on slope and we split the experiment into two blocks (aka, blocking): slope less than 10% and slope greater than or equal to 10%. Abe, Bob and Carl will be studied in both blocks and randomly allocated to sites in each block.
Block  Operator  
Slope < 10%:  Bob  Abe  Carl 
Slope ≥ 10%:  Carl  Bob  Abe 
What statistical analysis can be done? Similar to the monofactorial random design, basic calculations include productivity, costs and are calculated using standard units for the given machine, operator or system. Basic descriptive statistics and confidence intervals can also be determined. Treatment effects are tested using an Analysis of Variance (ANOVA) controlling for block error. What are the strengths and weaknesses of this design type?
Strengths  Weaknesses 


2.3.4 Multifactorial Block Design
A multifactorial block design involves testing two or more factors (Pretzsch, 2009). Block designs are used to reduce known systematic variation, (e.g. known changes in slope category, different shift times, etc.). This is done through the technique of blocking where treatments are grouped across the different categories (Clewer and Scarisbrick, 2001). Systematic bias is therefore controlled through a combination of the development of blocks (aka blocking) and the remaining bias is controlled through random treatment placement in the block (Pretzsch, 2009). Example of Design A study is designed to examine productivity of two operators (one of the treatment factors is operator), Abe and Bob, and the use of a cable skidder or grapple skidder (the second treatment factor). The site varies depending on gradient and the experiment is split into two blocks: gradient less than 10% and gradient greater than or equal to 10%. Abe and Bob operating each machine will be studied in both blocks and randomly allocated to sites in each block. What statistical analysis can be done? Basic calculations include productivity, costs and are calculated using standard units for the given machine, operator or system. Basic descriptive statistics and confidence intervals can also be determined. Treatment interactions as well as individual treatment effects are tested using factorial ANOVAs controlling for block effects. What are the strengths and weaknesses of this design type?
Strengths  Weaknesses 


2.3.5 Monofactorial Latin Square Design
A monofactorial Latin square design involves testing (or comparing) one factor or treatment (Pretzsch, 2009). The site however varies in two or more ways and this error is controlled through square (or rectangular) blocking (Clewer and Scarisbrick, 2001). Block designs are used to reduce known systematic variation, (e.g. known changes in slope category, different shift times, etc.). This is done through the technique of blocking where treatments are grouped across the different categories (Clewer and Scarisbrick, 2001). Blocks in a Latin Square design can be thought of as moving in rows and columns (Pretzsch, 2009). Example of Design A study is designed to compare the productivity of three operators, Abe, Bob and Carl (the treatment factor is therefore operator). The sites vary depending on slope and soil type. The experiment is therefore split into two rows (blocks) for slope less than 10% and slope greater than or equal 10%. The experiment will also be split into two columns (blocks) for clay type soil and sand type soil. Abe, Bob and Carl will be studied in both blocks and randomly allocated to sites in each block. What statistical analysis can be done? Similar to the monofactorial block design, basic calculations include productivity, costs and are calculated using standard units for the given machine, operator or system. Basic descriptive statistics and confidence intervals can also be determined. Treatment effects are tested using an Analysis of Variance (ANOVA) controlling for block error both for rows and columns. What are the strengths and weaknesses of this design type?
Strengths  Weaknesses 


2.3.6 Multifactorial Latin Square Design
A multifactorial Latin square design involves testing (or comparing) two or more factors or treatments (Pretzsch, 2009). The site however varies in two or more ways and this error is controlled through square (or rectangular) blocking. Block designs are used to reduce known systematic variation, (e.g. known changes in slope category, different shift times, etc.). This is done through the technique of blocking where treatments are grouped across the different categories (Clewer and Scarisbrick, 2001). In a Latin Square design, blocks can be thought of as moving in rows and columns (Pretzsch, 2009). Example of Design A study is designed to compare the productivity of three operators, Abe, Bob and Carl (the first treatment factor) and two skidder types, Cable and Grapple (the second treatment factor). The site varies in terms of gradient and average tree size. Two blocks will be formed for slope (less than 10% and greater than or equal to 10%) and two blocks for tree size (less than 1 m^{3} and greater than or equal to 1 m^{3}). Operators will be tested on both machines and operatorsmachine combinations will be randomly distributed across all blocks. The first letter in the site refers to the Operator (A,B and C) and the second letter refers to the machine type (C for cable and G for grapple). What statistical analysis can be done? Similar to the monofactorial block design, basic calculations include productivity, costs and are calculated using standard units for the given machine, operator or system. Basic descriptive statistics and confidence intervals can also be determined. Treatment interactions as well as individual treatment effects are tested using factorial ANOVAs controlling for block error in both rows and columns. Caution must be noted because, for this design, the number of replications can rapidly become very large (Clewer and Scarisbrick, 2001). For a solid design, every treatment must be replicated across all blocks, otherwise confounding effects can occur. Confounding can seriously diminish the strength of an experiment and should be approached with caution. What are the strengths of this design?
Strengths  Weaknesses 


2.3.7 Splitplot Designs
Split plot or split block designs are used in multifactorial experiments when one treatment can be applied on a large scale and the other treatment can be applied on a small scale (Clewer and Scarisbrick, 2001). Example of Design As an example, a study is designed to assess the effects of average tree volume (less than 1 m^{3} or greater than or equal to 1 m^{3}) and skidder type (cable or grapple) across three Pine species. Since tree volume is fixed by compartment, half the compartment is skidded with a cable skidder and the other half with a grapple skidder. The plot is therefore organised by average tree volume and then split by skidder type. What statistical analysis can be done? Beyond basic statistics and calculations, factorial ANOVAs would be used, although output of interactions and main effects becomes difficult to interpret. What are the strengths of this design?
Strengths  Weaknesses 


It is highly recommended that for studies which require this type of experimental design, a statistician should be consulted.
2.4 Modelling Studies
Similar to observation studies, modelling studies are done to observe machines, operators, or systems and create a production or cost model based on a series of input factors. These input factors must be measurable and preferably are continuous, meaning they are quantitative and within a range any number can exist. Examples of continuous variables include DBH, slope (%), speed, etc. Example of Design Develop a production model for a skidder in an operation. Inputs for this model include slope (%), cycle time, choking time, dechoking time, travel empty and loaded time, speed (loaded and unloaded), extraction distance etc. Some basic assumptions that need to be adhered to are homoscedacity and independence of data (see above). What statistical analysis can be done? Productivity and cost must be calculated in some way in order to develop the model. This can be done using regression methods, including multiple regression, or analysis of covariance. What are the strengths and weaknesses of this method?
Strengths  Weaknesses 

 May be difficult to control for variation from qualitative sources
2.5 Sample size calculations
It is essential that you have enough samples within your treatments and enough replications to allow for differences (or lack thereof) to be determinable. The difficulty that results is that in order to know the margin of error your sample will produce, you need to know the withintreatment variation (σ^{2}). The generic formula for sample size calculation is shown in Equation 1.
Where: n = sample size for study n’ = number of readings taken in the pilot study x = observed value Σ = sum of values (i.e.: sum of observed values)
2.5.2 Approximating Sample Size
3.0 Time Models
Time is a key element of production and is a crucial resource which must be managed. Several models are in use and work to describe how forestry activities use time. This standard will use the model and definitions proposed by the International Union of Forest Research Organisations (IUFRO). The IUFRO model (Figure 2) divides Total Time (TT) into NonWorkplace Time (NW) and Workplace Time (WP). Workplace time is further subdivided into NonWork Time (NT) and Work Time (WT). Work time is then divided into either Productive Work Time (PW) or Supportive Work Time (SW). Productive work time includes Main Work Time (MW) and Complementary Work Time (CW). Productive work time is where the work elements would be considered. Elements will be discussed further in Section 5. Figure 2: IUFRO time concepts structure (Björheden and Thompson, 1995) including abbreviations for time components. Supportive work time is further split into Preparatory Time (PT), Service Time (ST) and Ancillary Work Time (AW). From a time study perspective, the main objective is typically to determine the productive machine hours (PMH). These hours are what the IUFRO model refers to as Productive Work Time (PW). They are the portion of time where the machine, or operator, is engaged in their primary work function. For example, the productive machine hours for a chainsaw operator refer to the time he is actively felling trees, including the time to walk from tree to tree as this is fundamental to the felling process. In Section 5, detailed elements which demonstrate the machine/operations productive work cycle are described. From the time model, time can be divided up and used to calculate ratios which are essential for accurate costing. These ratios are: mechanical availability, machine utilisation and capacity utilisation. The ratios are calculated using time intervals developed from the IUFRO time concepts structure. These are detailed below (Table 2). Table 2: Description of time concepts used to calculate usage ratios.
3.1 Ratio Calculations
3.1.1 Mechanical Availability
Mechanical availability refers to the portion of the workplace time (WP) during which a machine is mechanically fit and able to conduct productive work (Björheden and Thompson, 1995). Availability is dependent on machine required maintenance, either preventative or otherwise (Pulkki, 2001). Equations 3 and 4 below detail the formulas for calculating mechanical availability. Both equations taken from Pulkki (2001).
3.1.2 Machine Utilisation
Machine utilisation refers to the portion of workplace time when a machine is used to conduct the function intended for the machine (Björheden and Thompson, 1995). It is dependent on the mechanical availability of the machine as well as on the effectiveness of the operating method (Pulkki, 2001). Equations 5 and 6 below detail the formulas for calculating machine utilisation). Both equations taken from Pulkki (2001).
3.1.3 Capacity Utilisation
Machine capacity utilisation refers to a measure of the extent of total time (TT) that the machine is used for work. This includes all delay times, supportive work time along with the actual productive work time (Pulkki, 2001). Equation 7 below details the formula for capacity utilisation. Equation taken from Pulkki (2001).
3.1.4 Visualisation of Time Concepts
Figure 3 below shows a diagram illustrating how the time concepts come together for use in ratio calculations. Figure 3: Time concepts visualisation for a machine operating over a 12 hour shift. 2 hours are spent on service time (ST), giving this particular machine a mechanical availability of 83%, 2 hours of shift are spent on operator rest time and another 2 hours are spent on other delays, such as answering personal cell phone calls. The mechanical utilisation of this operation is therefore 50%.
4.0 Time Study Techniques and Methodologies
Once an objective and appropriate experimental design have been decided, the study technique can be finalised. Study technique will be highly objective specific. There are six different types of time study techniques that are commonly used (Table 3). Each technique varies in its scope and duration. Cost of conducting a study also varies depending on how much time and resources it takes to conduct. Table 3: Comparison of the six time study techniques and their typical degree of scope and duration. Before discussing the individual study types, it is important to discuss delays. As shown above in the IUFRO model, Workplace time (WP) is divided into Work time (WT) and Nonwork time (NT). A delay is considered any time that is Nonwork time (NT). Delays can then be further classified depending on whether they are work related (WD) or Disturbance (DT). The literature tends to handle delays in differing ways, and the suggestion is often made that only delays greater than 15 minutes be recorded (Brown et al. 2010). We instead propose that any delay greater than 30 seconds be recorded and classified appropriately. Whether or not it is included in the analysis will depend on study length and sample size (a 20 minute delay on a oneday study may be unrepresentative but several 2 minute delays every day for a week could be) but it is felt it is important to at least have an understanding of where and when the delays occur. Whatever protocol is used, it is important that the person doing the study clearly states which route was followed so that comparison studies in the future become potentially possible.
4.1 Shift Level Study
A shift level study examines production of a machine, operator or system with the observational measurement being a fully completed work shift. This technique is generally used for longterm observation, monitoring or followup studies (Magagnotti and Spinelli, 2010).
4.1.1 Data acquisition methods
Data for a shift level study can be acquired either manually or automatically if the equipment is available. Manual shiftlevel studies involve giving a foreman or shift supervisor a sheet on which to record their team’s performance every shift. Specific data recorded should include:
 Shift start and end time
 Record of crew working
 Production in appropriate unit
 Job type
 Delays and causes of delays
 Fuel consumption
Some of this data may be collected automatically with onboard data logging software connected to appropriate sensors.
4.1.2 Advantages and Drawbacks
The major drawback to a shiftlevel study is that it requires ongoing data management, particularly if done manually, and that it lacks the finer elemental detail. Furthermore, shift supervisors need to support the study and understand their role in the study’s success is crucial. Nevertheless, it is a powerful tool and the analysis tends to be more straightforward, particularly when combined with a simpler experimental design.
4.2 Plot Level Study
A plot level study examines production of a machine operator or system with the observational unit being a fully completed plot. A plot can be designed specifically to meet the study’s objectives. An example of a plot would be 4 rows of 30 trees with consistent tree species, diameter, height and spacing. The unit therefore is a completed plot and time is cumulative for the entire plot (i.e. how long does it take Operator A to complete a plot versus Operator B).
4.2.1 Data acquisition methods
Data acquisition for a plot level study can be done manually or automatically depending on how the plot is defined and on the technology available. If a plot is smaller and contiguous with the next plot, it may not be possible to differentiate one plot from the next using data logging. If; however, plots are easy to separate then automatic acquisition is possible. For manual acquisition, the time study observer can time the duration of the plot and record the respective production figures. Other information to specifically include would be:
 Detailed definition of the plot
 Machine used, make and model
 Operator
 Species
4.2.2 Advantages and Drawbacks
The major drawback to a plot level study is it becomes difficult to compare a plot level study to other studies that do not use the same plot composition. Additionally, as timing focuses on the plot completion alone, delays and elemental data are not acquired. Furthermore, performance in a plot may be specific to the plot itself and may not be able to be applied outside. Nevertheless, the advantage of a plot level study is it is a very good way to quickly compare two very similar types of machinery or operators. Depending on the plot composition, it may also be easier to design an experiment for other study techniques.
4.3 Cycle Level Study
A cycle level study examines production on the cycle level and the observational unit is a completed cycle. A work cycle is defined as a sequence of tasks that perform a job or produce a unit of production (Kanawaty, 1992). A completed cycle can be anything from felling a tree to trucking a round trip with a load. Cycle level studies can be conducted manually or using automatic data acquisition depending on the objective of the study and the equipment available.
4.3.1 Data acquisition methods
For manual acquisition, an observer in field would record time consumed per cycle and note the relevant production figures. Delays should also be recorded and classified. Data loggers may also provide an alternative if appropriate sensors can be attached to the desired inputs (might be more difficult for chainsaws but feasible for forwarders).
4.3.2 Advantages and Drawbacks
The major drawback of a cycle level study is that it lacks the elemental detail of the work process. The advantages are that it provides a quick way of seeing the variability in the work process and allows delay information to be captured. It is less intensive than an elemental study. Overall, cycle level studies are not recommended.
4.4 Time and Production Count
One of the simplest techniques for time and work study is time and production count. The observation level is variable and can be anything from a cycle, series of cycles or a shift. Time and Production Counts are designed to be very quick and typically are done manually with an observer in the field over a few hours.
4.4.1 Data acquisition
Data collection is usually collected over a short period of time (i.e. few hours) and is done through recording productive time and production in the preferred unit (e.g. logs, volume, tons, etc.). Any delays should be recorded and excluded from productive time. By dividing production by time, one can find a quick estimate of performance. To calculate productivity, one of the following formulae should be used: Equations 8 and 9 taken from Brown et al. (2010). It is helpful to record comments on any special situation during study time (delays, work methods, etc.) as well as background information on the study conditions such as tree size, stocking, slope, etc.
4.4.2 Advantages and Drawbacks
The advantages of this technique are that it is quick and simple but the disadvantages are that the result only reflects the performance for a relatively short period of time and for a specific condition. In addition, it is difficult to identify inefficiencies because of the lack of detail.
4.5 Element Study
An element study breaks down the work cycle of a machine or system into individual functional steps called elements (Magagnotti and Spinelli, 2010). For the purposes of standardisation, elements have already been defined by machine type and are described indepth in section 5.0. An elemental study is typically conducted manually and tools can range from basic clipboard and stopwatch to complex handheld personal computers with detailed time study software to video recording. Particularly when individual elements are very short in duration, computer software and video recording can make capturing these elements easier.
4.5.1 Data acquisition
Elemental timing can be recorded using two different timing techniques: snap back timing or continuous timing (Magagnotti and Spinelli, 2010). In snapback timing, the clock is reset back to 0 at the end of every element. This can be done using the lap feature of a stopwatch. The major benefit of snap back timing is that recording the amount of time per element is very easy. A disadvantage is that it requires a watch that has a lap function or the observer to reset the clock every time which can increase the risk of timing mistakes. Continuous timing means that the time is recorded for every break point (transition between elements) and time per element is then calculated after the fact. Continuous timing is made simpler by the use of decimal time watches which convert minutes into decimal minutes allowing for simpler math. For elements that are extremely short, a handheld computer program which can change elements with one click can help to record very quick changes. The fastest elements though will require videotaping and element times are established through multiple replays after the fact (Magagnotti and Spinelli, 2010). Other data recorded includes:
 Any delay greater than 30 seconds and cause of delay
 Production unit
 Comments per cycle or element
4.5.2 Advantages and Drawbacks
The main advantage of an element study is the fine level of detail regarding the work process it provides. Element studies allow for greater understanding of the functional steps and can help directly pin down inefficiencies. The major drawback of elemental studies is they are time consuming and can become costly for acquiring large data sets. Experimental design has to be done to minimise replications and keep the overall number of observations feasible. Furthermore, element studies require the observer to be well versed in the element breaks and understand what they are specifically looking for.
4.6 Work Sampling (Instantaneous Observation, and/or Activity Sampling)
While not a true time study technique per say, work sampling is an important method of work measurement and is therefore recorded here. Similar to an element study, Work Sampling also records elementlevel data. Unlike time study; however, work sampling determines the relative frequency of the elements over the total time observed. During Work Sampling, a series of instantaneous readings of an activity are taken over a period of time. Ideally, the readings are not taken in time with the cycle as irregular sampling intervals.
4.6.1 Data acquisition
The observer collects data by sampling at either a fixed or random interval. Fixed intervals (e.g. two minutes) should be used in conditions when the duration of work activities are random. When the duration of activities are more systematic or when there is uncertainty regarding the duration of activities, sampling should be done at random intervals in order to avoid bias. With this technique, the relative times of work activities are determined by assuming that the percentage of observations recorded for each activity approximates the percentage of each activity within the total time. Each activity that occurs during each sampling interval is tallied and tallies are excluded from delays. To calculate the percentage of a particular activity within a work cycle, divide the total tally for that element by the total tally of the study.
4.6.2 Advantages and Drawbacks
Work sampling is a simple and inexpensive way to conduct time and work study, requiring only a wristwatch or stopwatch and a clipboard for equipment. No special training or expertise is needed to conduct a study using this technique and an observer may collect data on several pieces of equipment or operators at the same time. This technique provides a general time distribution and highlights efficiencies of a work cycle. However, it is difficult to apply to other conditions because of its lack of detail. A work sampling study is most effective when used for an operation where a number of activities are happening at once to complete a task. For example, a merchandising operation at roadside where multiple workers are cross cutting logs would be a good candidate for work sampling. Work sampling studies can also be used to assist in method determination or to help the data collector become familiar with a new machine or operation. One suggested use of work sampling is to couple it with an element study. The data collector performs work sampling for the first hour or so of the study and then switches to the element study. This first hour provides the benefit of allowing the workers to become accustomed to the data collector’s presence as well as provides a small work sampling dataset without any additional effort.
5.0 Machine Element Standardisation
A review was conducted to determine the machines commonly used in the South African forest industry. From this survey, elements in a normal machine cycle have been established. An element breaks down to a basic, functional step which can be measured throughout the duration of a normal work cycle. The pages below list the standardised elements and data collection requirements for commonly used harvesting machines in South Africa.
5.1 Standardised Element Lists by Machine
Please note the following:
 Any amount of additional detail can be added within each broad elements described below. A provision is that the timekeeper has to be able to record the duration of each additional element, all associated attributes are recorded and described, that the additional detail fits into the fixed elements as listed in the tables below, and that any additional breakpoints are properly described and recorded within the elements.
 If need be, one or more of the listed elements can be omitted for a study such as; e.g., with chainsaw felling the element “consideration”. Or it may be necessary to group “consideration “and “clear site”; or leave them out altogether. If two elements are grouped the recorder must make sure the breakpoints for start and end include the start for the first element and the end of the second element. However the timekeeper should consider the implication of this action before doing so as it can seriously affect the integrity of the study to be undertaken.
 The column “detail required” outlines the minimum required data and these range from time and distance to single tree dimensions. It is important in single tree operations that individual cycles’ match a specific tree data. If in doubt rather measure to smallest individual unit; e.g. single tree, log etc.
 It is important to become conversant with the “Time Model” described in section 3.0 for correct allocation of delays and systems operations. This is particularly important in the calculation of machine availability, machine utilisation and systems efficiency. Each delay must be adequately described and recorded.
Specific timestudy information required
 Time: Measure time in minutes and centiminutes – i.e. hundredths of a minute.
 Distance: Estimate the distance the operator moves from one task to the next. An approximate distance can be estimated in unthinned stands by using tree spacing. Otherwise the number of paces the operator takes can be used for good measure. If more accurate data is required use a tape measure. A GPS mounted to the operator could also provide a means of determining distances travelled between tasks.
 Single standing tree dimension: Number each tree to be felled in the study and pair this unique number to its associated dimensions. Also record other tree attributes; e.g., form etc. (refer to background information forms). Measure DBH (1.3m above ground level) and height of each tree. Use the Schumacher and Hall model (South African Forestry Handbook, 2012) to determine the volume of each tree. To convert to mass (tons) refer to the South African Forestry Handbook (2012 & 2000).
 Measuring equipment: Callipers (digital or manual), vertex (or other simple hypsometers – Suuntu) and tape measure or logging tape. For measuring methodology refer to South African Forestry Handbook (2000 & 2012).
 Individual log data:Record number of logs, and their dimensions (diameter – thin and thickend – and length) crosscut from each tree, if of interest.
 Refer to IUFRO Timemodels
Harvester:
Specific timestudy information required
 Time: Measure time in minutes and centiminutes – i.e. hundredths of a minute.
 Distance: Estimate the distance the machine moves between stops of tasks. An approximate distance can be estimated in unthinned stands by using tree spacing. Otherwise the number of rotations of the wheels/tracks can be used for good measure (mark a point on the wheel/track as reference point). Average speed for the move can be calculated as the quotient of distance and time for the move. GPS/OBC/CanBus system is a good alternative, if available.
 Boom movement:If machine has the ability to measure boom movement distance i.e. CanBus system, recover this data, otherwise exclude.
 Single tree attributes: Number each tree to be felled in the study and pair this unique number to its associated dimensions and record. Also record other tree attributes; e.g., form (refer background information forms). Measure DBH (1.3m above ground level) and height of each tree. Use the Schumacher and Hall model (South African Forestry Handbook, 2012) to determine the volume of each tree. To convert to mass (tons) refer to the South African Forestry Handbook (2012 & 2000).
 Individual log data: Record number of logs, and their dimensions (diameter – thin and thickend – and length) crosscut from each tree. Calculate log size (m^{3}/tons) from Huber, Samlain or Newton’s equations (South African Forestry Handbook 2012 & 2000). To convert to mass (tons) refer to the South African Forestry Handbook (2012 & 2000).
 Measuring equipment: Callipers (digital or manual), vertex (or other simple hypsometers – Suuntu) tape measure of logging tape. For measuring methodology refer to South African Forestry Handbook (2012 & 2000).
 Refer to IUFRO Timemodels
FellerBuncher :
Specific timestudy information required
 Time: Measure time in minutes and centiminutes – i.e. hundredths of a minute.
 Distance: Estimate the distance the machine moves. An approximate distance can be estimated in unthinned stands by using tree spacing. Otherwise the number of rotations of the wheels/tracks or another good approximation can be used for good measure (mark a point on the wheel/track as reference point). Average speed for the move can be calculated as the quotient of distance and time for the move. GPS/OBC/CanBus systems are good alternatives, if available.
 Tree data: As single tree dimensions (DBH specifically) do not affect time for felling operations greatly, single tree dimensions are not required provided the individual tree dimensions are relatively uniform throughout the work area. Use average tree volume/tons as a measure. To gain average tree volume, sample the compartment following the methodology outlined in the South African Forestry Handbook (2012 & 2000). To convert to mass (tons) refer to the South African Forestry Handbook (2012 & 2000). To convert to mass (tons) refer to the South African Forestry Handbook (2012 & 2000).
 Record the number of trees dumped: This will provide an estimate of the bunch size (number of trees and volume) for the extraction operation. Use average tree volume/tons as a measure. To convert to mass (tons) refer to the South African Forestry Handbook (2012 & 2000).
 Measuring equipment: Callipers (digital or manual) vertex (or other simple hypsometers – Suuntu) and tape measure or logging tape. For measuring methodology refer to South African Forestry Handbook (2012 & 2000).
 Refer to IUFRO Timemodels
Skidder/agricultural tractor with winch or drawbar (aframe or other):
Specific timestudy information required
 Time: Measure time in minutes and centiminutes – i.e. hundredths of a minute.
 Distance: Estimate the distance the machine moves along the forest road and/or from stump site to roadside landing (m). A close approximate distance can be estimated by staking the road/skid trail with pegs (e.g. 20 m to 50 m apart) as reference points. GPS/OBC/CanBus system is a good alternative, if available. Average speed for both the loaded and unloaded can be calculated as the quotient of distance and time for the move.
 Tree/load data: Record number of pieces contained in load dropped at the landing. To determine load size (m^{3}/tons) multiply average tree/log volume/tons with number of pieces. Calculate piece volume for logs and for longer lengths using Huber, Smalian or Newton’s equations (South African Forestry Handbook 2012 & 2000). Another option for longer lengths is to clearly mark DBH on the stem so that it is visible on arrival at roadside. Then record this DBH and the length of the tree and applying the Schumacher and Hall model (South African Forestry Handbook, 2012) for longer lengths or treelengths. To convert to mass (tons) refer to the South African Forestry Handbook (2012 & 2000). It may not be possible to measure each piece in high production operations. In this case determine a sample size (refer to protocol manual). Failing that a good estimate can be gained by measuring at least 30 pieces per day or per study and calculating volume/tons using the equations mentioned above.
 Measuring equipment: Callipers (digital or manual), vertex (or other simple hypsometers – Suuntu) and tape measure or logging tape. For measuring methodology refer to South African Forestry Handbook (2012 & 2000).
 Refer to IUFRO Timemodels
Forwarder :
Specific timestudy information required
 Time: Measure time in minutes and centiminutes – i.e. hundredths of a minute.
 Distance: Estimate the distance the machine moves along the machine trail (m). A close approximate distance can be estimated by staking the road with pegs (e.g., 20 m to 50 m apart) along the road as reference points. In this case a GPS is a good alternative, if available. Average speed for the move can be calculated as the quotient of distance and time for the move. GPS/OBC/CanBus system is a good alternative, if available. Average speed for both the loaded and unloaded can be calculated as the quotient of distance and time for the move.
 Tree data Record number of pieces contained in each grapple load. To determine load size (m^{3}/tons) use an estimation of average tree/log volume/tons by using Huber, Smalian or Newton’s equations (South African Forestry Handbook 2012 & 2000). Also record the number of grapple loads to complete the loading of the forwarder. Total load size can be estimated by multiplying the total number of logs in the full load with the average piece size. To convert to mass (tons) refer to the South African Forestry Handbook (2012 & 2000).
 It may however not be possible to measure each piece loaded that makes up the total load. In this case a sample of logs must be measured to determine an average log size and used throughout the study (if piece size remains uniform). Determine the minimum sample size needed using the sample size calculator outlined in the Protocol manual. Failing that a good estimate can be gained by measuring at least 30 pieces per day or per study and calculating volume/tons using the equations mentioned above.
 Measuring equipment: Callipers (digital or manual), vertex (or other simple hypsometers – Suuntu) tape measure of logging tape. For measuring methodology refer to South African Forestry Handbook (2012 & 2000).
 Refer to IUFRO Timemodels
Loader (either tracked or wheeled):
Specific timestudy information required
 Time: Measure time in minutes and centiminutes – i.e. hundredths of a minute.
 Distance: Estimate the distance the machine moves along the machine trail (m). A close approximate distance can be estimated by staking the road with pegs (e.g. 20 m to 50 m apart) along the road as reference points. In this case a GPS is a good alternative, if available. Average speed for the move can be calculated as the quotient of distance and time for the move. GPS/OBC/CanBus system is a good alternative, if available. Average speed for both the loaded and unloaded can be calculated as the quotient of distance and time for the move.
 Tree data Record number of pieces contained in each grapple load. To determine grapple load size (m^{3}/tons) use an estimation of average tree/log volume/tons by applying Huber, Smalian or Newton’s equations (South African Forestry Handbook 2012 & 2000). Also record the number of grapple loads required to complete the loading of the vehicle. Total load can be estimated by multiplying the total number of logs in the full load with the average piece size. To convert to mass (tons) refer to the South African Forestry Handbook (2012 & 2000).
 Measuring equipment: Callipers (digital or manual), vertex (or other simple hypsometers – Suuntu) and tape measure or logging tape. For measuring methodology refer to South African Forestry Handbook (2012 & 2000).
Refer to IUFRO Timemodels
Processor:
Specific timestudy information required
 Time: Measure time in minutes and centiminutes – i.e. hundredths of a minute.
 Distance: Estimate the distance the machine moves along the machine trail (m). A close approximate distance can be estimated by staking the road with pegs (e.g., 20 m to 50 m apart) along the road as reference points. In this case a GPS is a good alternative, if available. Average speed for the move can be calculated as the quotient of distance and time for the move. GPS/OBC/CamBus system is a good alternative, if available. Average speed for both the loaded and unloaded can be calculated as the quotient of distance and time for the move.
 Boom movements: If the machine has the ability to measure boom movement distance, recover this data, i.e. CanBus, OBC etc., otherwise exclude.
 Tree data: Record number of logs produced from each processing event. Do not separate debarking from crosscutting as it is very difficult to define each operation separately. If possible record passes with eucalyptus debarking if finite
 Measuring equipment: Callipers (digital or manual), vertex (or other simple hypsometers – Suuntu) and tape measure of logging tape. For measuring methodology refer to South African Forestry Handbook (2012 & 2000).
 Refer to IUFRO Timemodels
Truck (timber transport):
Specific timestudy information required
 Time: Measure time in minutes and centiminutes – i.e. hundredths of a minute.
 Distance: Record distance travelled (km) either from speedometer of by means of GPS data.
 Road class: Road class can be added and recorded if desired; otherwise, note on background information form.
 Tree data:Record load size by recording then number of logs loaded multiplied with average tree/log volume/tons. Determine log volume/tons using Huber, Smalian or Newton’s equations (South African Forestry Handbook 2012 & 2000). To convert to mass (tons) refer to the South African Forestry Handbook (2012 & 2000). An alternative to determine load size is to use recorded by the truck scales.
 Measuring equipment: Callipers (digital or manual), vertex (or other simple hypsometers – Suuntu) and tape measure or logging tape. A GPS is another useful tool for truck distance measurement, particularly when longer distances are being studied. For measuring methodology refer to South African Forestry Handbook (2012 & 2000).
 Refer to IUFRO Timemodels
Yarder:
Specific timestudy information required
 Time: Measure time in minutes and centiminutes – i.e. hundredths of a minute.
 Distances: Estimate the distance the carriage moves on the skyline loaded (m). A close approximate distance can be estimated by staking the corridor with pegs (e.g., 20 m to 50 m apart) as reference points. GPS can also be used. Average speed of the carriage return process can be calculated as the quotient of distance and time for the move.
 Tree data: Record number of pieces contained in load dropped at the landing. To determine load size (m^{3}/tons) use an estimation of average tree volume/tons using Schumacher and Hall model (South African Forestry Handbook, 2012) or Huber, Smalian or Newton’s equations for logs (South African Forestry Handbook 2012 & 2000). To convert to mass (tons) refer to the South African Forestry Handbook (2012 & 2000). It may not be possible to measure each piece in high production operations. In this case determine the minimum sample size needed to produce an acceptable estimate (refer to protocol manual for sample size calculator). Failing that a good estimate can be gained by measuring at least 30 pieces per day or per study and calculating volume/tons using the equations mentioned above
 Measuring equipment: Callipers (digital or manual), vertex (or other simple hypsometers – Suuntu) and tape measure or logging tape. For measuring methodology refer to South African Forestry Handbook (2012 & 2000).
Refer to IUFRO Timemodels
Mulchers and Destumpers:
Specific timestudy information required
 Time: Measure time in minutes and centiminutes – i.e. hundredths of a minute.
 Distances: Use a measuring wheel to measure distance. This can be done afterwards if a starting point, the rows and the end point are marked. Distance between stumps can be calculated using the compartment spacing.
 Stump data: single stump dimensions (height and diameter) are not required provided the individual stump dimensions are relatively uniform throughout the work area. Use average stump dimension as a measure.
 Record the number of stumps treated
 Refer to IUFRO Timemodels
5.2. Userdefined elements
The user is strongly encouraged to use these predefined elements for both convenience and the purposes of industry standardisation; however, in certain cases, developing new elements may be required (e.g. the user is examining a machine which is not on the list below). All elements are basic, functional steps that occur during the work process, whether they contribute to the successful completion of work or not (delays). When defining elements, a key consideration is defining element breakpoints. Breakpoints refer to the exact start and exact end time of an element. For example, a refuelling time element for a chainsaw begins when the saw stops due to lack of fuel or fuel top up and resumes when the saw starts to continue the operation. Elements also need to have defined measurement standards. This may be just the length of time the element takes to complete but it may also have other data requirements, such as the volume of load. When new elements are used, the user is kindly asked to define these steps and forward this information along to FESA in order to continue improving this protocol.
6.0 Statistical Analysis
This section is still in progress.
7.0 References
Barnes RM. 1963. Motion and Time Study – Design and Measurement of Work. 5^{th} Edition. London: John Wiley & Sons Inc. Björheden R, Thompson MA. 1995. An International Nomeclature for Forest Work Study. In DB Field (Ed.), Proceedings of IUFRO 1995 S3:04 subject area: 20^{th} World Congress (pp. 190215). Tampere, Finland: IUFRO. Bredenkamp BV, Upfold SJ. 2012. South African Forestry Handbook 5^{th} Ed. Southern African Institute of Forestry. Brown M, Acuna M, Strandgard M, Walsh D. 2010. Machine evaluation toolbox. Hobart, Tasmania: Cooperative Research Centre for Forestry Australia. Clewer AG, Scarisbrick DH. 2001. Practical Statistics and Experimental Design for Plant and Crop Science. London: John Wiley & Sons. 332 pp. Cochran WG. 1977. Sampling Techniques (3^{rd} edn) .New York: John Wiley & Sons.428 pp. Kanawaty G (Ed.). 1992. Introduction to Work Study (4^{th} Edn.). Geneva: International Labour Organisation. Magagnotti N, Spinelli R. (Eds.) 2010. Good Practice Guidelines for Biomass Production Studies. Sesto Fiorentino: CNR IVALSA. Milton JS, Arnold JC. 1999. Introduction to probability and statistics: principles and applications for engineering and the computing sciences (2^{nd} edn). New York: McGrawHill. Ott RL. 1993. An introduction to statistical methods and data analysis (4^{th} edn). Belmont: Wadsworth Publishing Company. 1056 pp. Pretzsch H. 2009. Forest Dynamics, Growth and Yield. Berlin: SpringVerlag. 663 pp. Pulkki RE. 2001. Forest Harvesting I: On the Procurement of Wood with Emphasis on Boreal and Great Lakes St. Lawrence Forest Regions. 156 pp.