INFLUENCING FACTORS ANALYSIS TO DETERMINE STANDARD TIME OF OPERATORS IN A GAUGE CALIBRATION PROCESS

Karen Botelho Faria

Universidade Candido Mendes, Brazil

E-mail: karenbotelhofaria123@gmail.com

Ana Carla de Souza Gomes dos Santos

Instituto Federal do Rio de Janeiro, Brazil

E-mail: anacarla.engenharia@gmail.com

Layanne Andrade Mendonça

Instituto Federal Fluminense, Brazil

E-mail: layanne.mendoca@iff.edu.br

Augusto da Cunha Reis

CEFET, Brazil

E-mail: augusto@aaa.puc-rio.br

Sérgio Augusto Faria Salles

Instituto Federal Fluminense, Brazil

E-mail: safsalles@hotmail.com

Submission: 29/03/2018

Accept: 29/03/2018

ABSTRACT

Companies seek to improve service level, aiming to meet the most demanding customers that enable it survives in a complex and competitive marketplace. The article presents a study in a company of repair and maintenance of industrial equipment and instruments, located in Campos dos Goytacazes (RJ), aiming to identify the factors that influence the execution time of the operators' activities in the manometer calibration process. For this scope, the study of times and movements and statistical analyzes, such as Linear Regression and ANOVA, were applied. The survey revealed that operators who have completed their technical courses in mechanics perform the activities in standard time greater than those who have not yet completed and that working on board and being married do not change the execution time. It also revealed that employees of lower age and longer experience are preferable because of shorter operating times and that the age was statistically significant, revealing that the most appropriate age to have a shorter time is approximately 27 years.

Keywords: time-motion study; productivity; regression; ANOVA

1. INTRODUCTION

Service sector has faced a decrease of 3.2% on volume of services rendered in July 2017 in comparison with the same month a year ago and a decrease of 0.8% in comparison with its previous month (IBGE, 2017). As a result, companies at a high level of competitiveness are aiming to meet customer requirements and remain survivor in the marketplace.

Some standardization tools such as time and motion study (TMS), first proposed by Frederick Taylor, provides competitiveness for a company and is still currently applied in several organizations (NOVASKI; SUGAI, 2002; SILVA; GUIMARÃES; VERALDO JÚNIOR, 2014; SIMÕES; FENNER; ESPERANCINI, 2014). This study enables reduction of the cycle time, best control over the process, standardization of the way of work and an increase in production capacity in view of encouraging competitive differentiation.

According to Figueiredo, Oliveira and Santos (2011), time and motion study is applied in order to eliminate unnecessary efforts when performing a given operation by establishing a certain way to perform it and thus obtain methods to assure an overall improving of the process.

By applying TMS it is also possible to identify which operators are performing the activities above or below the value established by the company (SANTOS et al., 2015). However, it is extremely important to find out what are the factors that influence this performance time. This sort of analysis is possible through statistical techniques, such as ANOVA (for qualitative data) and Linear Regression (for quantitative data).

Therefore, the aim of this article is to provide an application of time and motion studies to identify the factors that influence the execution time of the activities performed by the operators. The study was carried out in a company that provides equipment repair, maintenance and repair service according to NR-13, located in Campos dos Goytacazes, specifically in the calibrating gauge process.

This paper is structured in five sections as follows. Section 1 is the introductory one in which the article is contextualized. Section 2 reviews time and motion study key concepts in which the study is based on. In Section 3, the methodology proposed for the study is described. The results and subsequent discussions are provided in Section 4 and, finally, the paper is concluded in Section 5.

2. TIME AND MOTION STUDY

According to Chiavenato (2014) and Santos et al. (2015), time and motion studies were first suggested in 1881 by Frederick Taylor while working at the Midvale Steel Company. After becoming chief operating engineer, he decided to change the management system in a way that the interests of both factory floor workers and high managers would not conflict.

Soon, he realized that the great obstacle was the unfair workload given to the workforce and got permission to apply a certain amount of money in a scientific study in order to determine the best production time. He chose two healthy and efficient workers and aimed to find out what a “fair day’s work” meant for an efficient worker; i.e., the best way a man could do his job year after year successfully.

Thus, Taylor discovered that the energy spent during the execution of heavy jobs were correlated with its rest periods, frequencies and durations and also established that there would always be a faster method and a better instrument to be found and improved by means of scientific analysis and time/motion study.

Barnes (2004) indicates that time and motion study is used to determine the time required for a properly trained and qualified operator perform a specific task as well as its most appropriate execution method. Moreover, he states that time and motion study has the following purposes: (i) to find the most appropriate method for performing a task, (ii) standardize this method, (iii) determine the time taken by a given operator to perform the task at a normal pace, and (iv) guide and train the operator in the chosen method.

According to Peinado and Graeml (2007), time and motion studies are not only applied to define a better way of perform a task, but also to determine the productive capacity of an organization and elaborate production and cost reduction programs, among other applications.

For the work to be executed in a better and more efficient way, it must be analyzed under a specific method. The first step is to breakdown the operation into smaller and observable elements. Then, the number of repetitions is defined and the operator’s time for each element is measured and recorded using a stop watch. This analysis is used to determine the performance time of the overall productive process, each task and each operator (CHIAVENATO, 2014; SANTOS et al., 2015). The observed time is relevant if the time to be measured exceeds five seconds. Thus, with the assistance of a stop watch, each step is timed several times so that the arithmetic mean can be calculated (JACOBS; CHASE, 2012).

After that, it is necessary to calculate the number of cycles (N), so the number of elements collected in the study is valid and it can arrive at an accurate average time. In other words, the formula determines the minimum number of repetitions. A number greater than or equal to the value provided by the formula is statistically feasible and the average time found can be used. Equation 1 determines the minimum sample number:

(1)

Where:

N: Number of cycles;

Z: Standard normal distribution coefficient;

R: Sample’s range;

Er: Relative error;

: Coefficient related to number of times previously observed;

: Arithmetic mean of observed times.

It is acceptable to use a confidence interval between 90 and 95% to determine the standard normal distribution coefficient (Z), and the relative error (Er) with a margin between 5 and 10%. The values of Z and d₂ are set out in Tables 1 and 2 respectively:

Table 1: Standard normal distribution coefficient (Z)

Probability	90%	91%	92%	93%	94%	95%	96%	97%	98%	99%
Z	1.65	1.70	1.75	1.81	1.88	1.96	2.05	2.17	2.33	2.58

Source: Adapted from Peinado and Graeml (2007)

Table 2: Coefficient related to number of times previously observed

N	2	3	5	5	6	7	8	9	10
d2	1.128	1.693	2.059	2.326	2.534	2.704	2.847	2.970	3.078

Source: Adapted from Peinado and Graeml (2007)

The next step is to calculate the performance rating of each operator using the average time of the productive process, called observed time (OT), and the average time of each operator, called normal time (NT). OT is equivalent to a normal rate of 100%. Therefore, the speed of each operator is calculated by the Equation 2 below:

(2)

Where:

NT: Normal Time;

OT: Observed Time;

PR: Performance rating of each operator.

There are some reasons that may lead the worker to be at an above-standard rate, for example: fatigue, lack of practice, family problems, lack of motivation, company time, inadequate training or lack of training, age, work environment and so on. Thus, the performance rating calculation reveals whether the operator is: slow (above 100%); fast (below 100%); or in the ideal time (equal to 100%).

According to Martins and Laugeni (2015), it is impossible for an operator to work all day without interruption or break. Thus, breaks during working hours are granted to workers so that they can meet their physiological and personal needs, thereby minimizing their fatigue and even resulting in increased productivity when they return to their jobs. Allowances or tolerance (p) is made by means of the sum of time that the company grants its employees, including time for delays and alimentation, due to the workload according to Equation 3 below:

(3)

After calculating the allowance time (p), the tolerance factor (TF) is calculated by Equation 4 below:

(4)

Where:

TF: Tolerance factor;

p: Allowances.

The tolerance factor (TF) obtained is added to the OT and NT of each operator, resulting in the standard time (ST), according to Equations 5 and 6 below:

(5)

(6)

Where

ST= Standard time;

NT= Normal Time;

TF= Tolerance factor;

OT= Observed Time.

These two calculations differ from each other since the calculation of the ST of the operator is the multiplication of the NT by the TF. The standard time of the company is calculated by multiplying the OT by the TF.

2.1. Linear regression and time and motion study

According to Golberg and Cho (2004), the regression analysis consists of obtaining an equation that tries to explain the relationship among variables. In order to establish an equation that represents this analysis, it is necessary to draw a graph, called a scatter diagram, in order to verify how the values of the criterion variable (Y) depends on the variation of the independent variables (X).

It is observed whether the points in the diagram drawn fit the model line. For the study to be consistent, the model chosen must match what happens in practice, containing only the factors that are relevant to the explanation of the phenomenon. The linear regression models can be, among others, first degree or second degree, as shown in Equations 7 and 8, respectively:

(7)

Where:

is the value observed at the i-th level of variable X;

is the regression coefficient;

is the constant of the regression;

is the i-th level of variable X;

is the error associated with the difference between the observed value and the estimated value for the same level i of the variable X.

This model is a first-degree polynomial function whose graph is a straight line. The positive coefficient indicates that the function is increasing, i.e., increasing the levels of the variable X studied, the response variable is also increased. However, if the regression coefficient is negative, it means the lower the factor level, the higher the response variable according to Equation 8 below:

(8)

Where:

is the value observed at the i-th level of variable X;

and are the regression coefficients;

is the regression constant;

is the i-th level of variable X;

is the i-th level of variable X squared;

is the associated error that means the difference between the observed value and the estimated value for the same level i of the variable X.

The adjustment of the regression by a second-degree model occurs when the data do not behave around a line but a parabolic curve.

The process of identifying the optimal level of the factor that generates a greater or lesser value of the dependent variable is called optimization. A first-degree model with a positive coefficient, adjusted for a given range of factor levels, has a higher response value for the higher factor level of this interval. If the regression coefficient is negative, the highest value of the response variable is obtained using the lowest level of the factor studied.

A second-degree model, whose graph is a parabola, will have a maximum value if the coefficient associated with the quadratic term is negative. Otherwise, the adjusted equation will have a minimum value. The optimization process of this model occurs through the identification of the value of the level of the factor that maximizes or minimizes the dependent variable, denoted and calculated by Equation 9:

(9)

To determine the maximum or minimum value of the dependent variable, the factor must be simply replaced in the adjusted equation.

The coefficient of determination is a fundamental parameter in determining the best adjusted model. This can have a value between 0 and 1 and the closer it is to 1, the better the model explains the variation of the data. If is close to 0, it implies that there are other causes that did not enter the model, but should be studied or that another model should be adjusted.

By associating the application of linear regression to time and motion study, it is possible to identify which are the factors that affect the execution time of the operators. Each operator has an average time to perform the task, however when calculating the standard time of the operator and comparing it to the standard time of the company, it is concluded that some operators are slow and others are faster.

According to Peinado and Graeml (2007), the determination of the performance rating or rhythm of the worker is the most important and difficult detail of time study. Velocity was previously calculated subjectively by a specialist. Barnes (2004) confirms that the professional assigned to the timing when recording the survey data was responsible to evaluate the performance rate of each operator in relation to its normal speed reference in a personal way, not taking into account the variables that may affect the performance of the operator, such as family problems, execution time of the task to be studied, adequate training, pressure of superiors, among others.

3. RESEARCH METHODOLOGY

This section describes the research method adopted in this paper so that all steps can be replicated in future studies.

3.1. Data collection

As highlighted by Amato and Reis (2016) a case study methodology allows an investigation of current events in a practical and holistic way in which the researcher has little or no control over the object in question. The research is classified according to its nature, since it aimed to generate new knowledge from a local practical application. To achieve this objective, this research is a case study which was conducted taking into account the unit of analysis that is the calibration process of manometers.

Regarding the approach, the research is characterized as both quantitative and qualitative, since it covered variables that were translated into numbers and has subjective characteristics. From the point of view of the objective, the research is classified as explanatory, because "it aims to identify the factors that determine or contribute to the occurrence of the phenomena" (SILVA; MENEZES, 2005).

Initially, a research was done on the topics that this article possesses so that the reader could clearly understand its purpose. After, the study object was chosen. The chosen process was the calibration of manometers based on the great representation that this operation has in the company's revenue.

Then, a survey, a division of the activities that compose the calibration of a manometer and the selection of five operators were made. This selection was made at random, according to company availability. Therefore, eight visits were made in loco, being five in the month of September and three in the month of November of 2016, for the accomplishment of the timekeeping.

3.2. Data processing and analysis

The stages related to time and motion study were performed: operators’ activities time were recorded using a stopwatch and a drawing board; observed time (OT) were calculated to each small operation and to the overall production process; normal time (NT) and performance rate of each operator were defined; tolerance factor (FT) was determined; and finally, standard time (TP) was calculated. In view of the variability of the operators' times, it became impossible to calculate the number of cycles.

The data were submitted to statistical analysis aiming to identify the factors, or independent variables that may interfere with the standard time response variable of the operators. Assuming normality, data independence and homogeneity of variance, an Analysis of Variance (ANOVA) was performed with qualitative levels factors. In this situation, it was intended to know whether the factors such as completed technical mechanical course, children, marital status and/or work situation, influence operators’ performance. Each factor has two qualitative levels, described in the table below:

Table 3: Qualitative levels

	Factors
	Married		Children		Work situation		Education
Levels	Yes	No	Yes	No	Onshore	Offshore	Complete	Incomplete

The hypotheses of interest in an ANOVA are as follows:

In this formula, μ1 stands for the mean of the "yes" or "completed" level of the factor under analysis and μ2, stands for "no" or "not completed" level. If the p-value calculated is lower than the established level of significance (α), the hypothesis of equal means is rejected; thus, a difference between factor levels is detected. This means that changing the "yes" to "no" or "completed" to "not completed" level of the factor under analysis causes change in the standard time.

The results were interpreted comparing the averages of the studied levels. Those with lower values were recommended, indicating a shorter activity operating time.

For the quantitative analysis, the standard time was used as dependent variable as a function of the factors of experience, age and quantity of children of the operators. The table below presents the factors and their respective levels used in the analysis:

Table 4: Experience time, age and number of children

Factors
Levels	Experience time (months)	Age (years)	Number of children
	18 24 30 36 60	20 23 32 33 40	0 1 2

Correlation analysis was used to analyze the strength of the relationship between the factors and the standard time. Subsequently, a first degree or second degree linear regression adjustment was made. For the factors that were statistically significant (p-value <α) in the fit of the first-degree model, an adjustment for a second-degree model was then made, since they could have a parabolic behavior. The model with highest coefficient of determination (R²) was the one defined as the best represent of the situation, indicating that this model explains better the variation of the data collected.

For the adjustment of the first-degree and quadratic linear equations, the models represented in Equations 7 and 8 were used respectively. For the second-degree adjustments, the optimization was done by calculating the value of the vertex of the parabola (