PRODUCTIVITY INDICATORS AS DECISION SUPPORT INSTRUMENTS IN STORAGE OPERATIONS

Since 1990, Brazilian Government has been promoting the implementation of Quality and Productivity improvement programs. In this environment, it is important to study the development of productivity measurement, since it supports the application of these programs.Far from being simple, productivity measurement must consider the influence of external factors which may affect the real productivity achieved. If this is not done, it may lead to wrong conclusions about labor performance.

In this paper, a set of performance measures for the loading activity in a Lubricant Factory has been defined. This activity is one of the most important for the warehousing function in any company. Finally, an analysis has been done to show the importance of considering external factors that may influence productivity.

Keywords: measurement; productivity; quality; shipping; warehousing

1. INTRODUCTION

The activity of loading boxes and buckets onto trucks at a lubricants factory has a very relevant influence on the company's relationship with the carriers that provide services, as well as on the level of service provided to customers (delivery time, number of breakdowns, etc.).

Therefore, the initial objective was translated into the definition of productivity measures that encompassed the activity in question. Traditionally, the main objective of establishing a productivity measurement system for a given activity is to provide the necessary information to improve productivity, people allocation and work performance. It becomes important to monitor current results, compare them with historical values, measure operator performance, monitor progress made and even help the employee evaluation process.

It is worth emphasizing the difference between performance (performance) and productivity (output versus input). Performance is a measure of an individual's level of effort and ability, it is the ratio of actual output to standard output. Productivity represents the ratio between generated outputs (output) and consumed resources (input).

The definition of performance standards is very useful for maintaining the desired level of productivity. This definition can be made using different techniques, such as studying time and motion, calculating estimates or productivity targets or historical averages.

For each function, a large number of work elements can be evaluated, each with its own unit of measurement. Therefore, appropriate measures should be selected, that is, covering the entire analyzed process. Measures can provide very useful information to those who want to identify sources of problems or reasons for successes. In addition, they can help the development of control systems that allow safe decision-making and facilitate the implementation of quality programs. Measuring productivity therefore means measuring the efficiency of a process. As was said earlier:

Productivity = Output Produced / Input Consumed

where Output is a measure of the amount of work performed on an activity (for example, number of items carried or weight carried) and Input is a measure of the resource consumed to perform the work (for example, man-hours or machine-hours).

In addition to the concepts of productivity and performance, there is the concept of utilization, which represents the fraction of input consumed in a given activity in relation to the total amount available for use.

Productivity can be increased in three ways: re-engineering the process itself, improving the use of resources and increasing performance through targets or other incentives.

Currently, it is possible to implement measurement systems in which complex calculations and large databases are necessary since we have computational resources that help in these tasks. The main task, therefore, is to decide what should be measured, what data to use to develop measures important for decision-making, and how to present such measures in a way that people can understand and interpret them.

To decide what to measure, a company must ask four questions: “What are you trying to accomplish?”; “What represents a good measure of this?”; “What mathematical formula calculates this?”; “What are the data sources?”. By following this logical sequence, it is possible to avoid getting lost in the middle of excessive data, being able to assure that the measurements are reaching the correct target.

In addition, the set of measures must have certain characteristics that guarantee its effectiveness. Such characteristics can be summarized as: validity (Does the measure provide tracking of customers' true needs or actual productivity?); coverage (does the measure or group of measures cover all relevant factors?); comparability (can the measure be compared over time and in different locations?); coverage (are all sources that generate an output covered by the measures?); usefulness (does the measure serve to guide actions?); compatibility (is the measure compatible with existing data and the flow of information?); cost/benefit (what are the trade-offs between the cost of obtaining the measure and the potential benefits to be achieved?).

2. SYSTEM DESCRIPTION

As seen previously, the activity studied in this work consists of loading trucks in the warehouses with buckets and boxes (dry cargo) of a lubricant factory. In this item, a brief description of the considered system is made.

The truck, truck or trailer, after being ready to load, pulls up to a dock in the warehouse for cans and buckets. The cargo is moved by two forklifts and manually depalletized inside the truck. After finishing loading at the bucket warehouse, the truck leaves the dock and goes to another warehouse, where the boxes are loaded. Once at the dock, the cargo is loaded onto the truck in the same way as in the bucket warehouse. It should be remembered that a truck has a capacity of up to 15.000 kg and a trailer carries up to 30.000 kg. In the implementation stage of the measures carried out in this study, the total loading time was calculated as the time elapsed between the beginning of loading in the bucket warehouse and the end of loading in the box warehouse.

3. DEFINITION OF THE SET OF MEASURES

From the previously explored concepts and the information obtained about the operation, some productivity measures appropriate to the shipping activities were selected. These measures are classified into the following groups: work, facilities and equipment.

Work Indexes

• Loading rate in kilograms per hour: represents the average weight (in Kg) loaded on each truck within an hour. The calculation is made by dividing the total weight loaded on each truck by the difference between the final and initial loading times of this truck.
• Vehicles loaded per hour (in trailers/hour or trucks/hour): represents the fraction of vehicle (truck or trailer) that is loaded in an hour. It is calculated by dividing the loading rate by the average load of the considered truck.
• Relative performance between different types of truck: represents the fraction of a truck loaded in time corresponding to the total loading time of another truck of a different type. The calculation is the ratio between the measures of vehicles loaded per hour of different types of truck.

Installation Indexes

• Daily productivity per dock (in Kg/dock.day): represents the daily loading rate at a dock, that is, how many kilograms were moved at each dock in one day. Its calculation is done by dividing the total loaded in a day (in Kg) by the number of docks in operation on that day.
• Dock turnover (in Vehicle/dock.day): represents the average number of vehicles that pull up to load at each dock in a day. It is calculated by dividing the number of vehicles loaded in a day by the number of docks in operation on that day.

Equipment Index

• Forklift productivity (kg/hour.forklift): represents the weight (in Kg) moved by each forklift within an hour. It is calculated by dividing the daily total loaded (in Kg) by the product between the number of forklift trucks and the total hours of use per day.

These were, then, the productivity measures suggested for the loading operation of dry cargo trucks in this lubricant factory. In the next item, some values ​​obtained for each of these measures are exposed, for example, from the collection of a data set for the month of February 1995.
4. IMPLEMENTATION OF MEASURES

For the selected Work, Facilities and Equipment Indexes, the average values ​​below were obtained from a sample of 127 vehicles (4 days). In this sample, 58% of the vehicles were trailers and 42% were trucks.

Work Indexes

loading rate

For vehicles in general 3818,30 Kg/h
For trucks 2942,51 Kg/h
For trailers 4496,32 Kg/h

Vehicles loaded per hour

For vehicles in general 0,21 vehicles/hour
For trucks 0,26 trucks/hour
For carts 0,19 carts/hour

Relative Performance

TCtruck = 0,70 TCtruck

where TC = loading time (sample average)

Installation Indexes

Daily productivity per dock

For vehicles in general 40267,68 Kg/dock.day

Dock Tour

For vehicles in general 2,12 vehicles/dock.day
For trucks 0,88 trucks/dock.day
For trailers 1,23 trailers/dock.day

Equipment Index

Productivity of Forklifts
8628.79 Kg/hour.forklift

As already mentioned, the values ​​presented above represent only the average values, obtained in a short period of time (only four days), for the suggested measures. For an effective use of the productivity measurement system, it is necessary to monitor the values ​​of the indexes over time, comparing them with historical values ​​or goals to be achieved. Comparing measurements before and after implementing some change in the operating system can give a comprehensive picture of the effectiveness of that change.

To exemplify and show how the graphic visualization facilitates the analysis of the measures, the load rate work index graph is shown below. Each point represents the daily average for the period from February 17 to April 01, 1995.

Graph 1 - Loading Rate in kg/h (Trucks and Trailers)
The averages obtained for productivity of trucks and trailers were 4831.78 Kg/h and 6067.80 Kg/h, respectively.

It can be noticed that the loading productivity in trailers is almost always higher than in trucks, although the latter vehicle presents increases in productivity proportionally greater than those of trailers. Anyway, as the trailer holds on average twice the volume of cargo than a truck, but has a higher loading rate (productivity), we conclude that there is a gain in scale in loading, that is: despite the trailer being twice bigger than a truck, the boarding time is not necessarily twice as long. This gain can be easily explained by the fact that the average value previously obtained for the Relative Performance index between trailers and trucks was equal to 70%, that is, the time required to load a truck is equivalent to the time required to load 70% of a trailer (on average), as opposed to the expected 50% if there were no gains in scale.

Thus, throughout the analysis of the collected data and from information provided by the people responsible for loading at the lubricant factory, an interdependence between some factors that influence loading productivity was evidenced, such as the type of truck (trailer or truck) and the type of load (fractional or unitized). It should be clarified that a fractional load is made up of many different items, whereas a unitized load is made up of a few different items. Therefore, it was decided to analyze whether there really is an influence of such factors on the productivity achieved and, if so, to eliminate erroneous interpretations about the variations in productivity values. For example, if the average value of labor productivity for a month is lower than that observed in the previous month, this does not necessarily mean that the quality of the operators' work has declined. There may have been a greater occurrence of trucks with fractional loads, or the percentage of trucks may have been considerably higher than that of trailers.

5. EXPERIMENT PROJECT

The main objective of this step was to obtain information about the sensitivity of the system to variations in two factors, which seemed to have a stronger influence on its performance. These factors are the type of truck (trailer or truck) and the type of load (fractional or unitized).

First, it was decided to verify whether these factors really influenced the productivity achieved. For this, a sample of 668 shipments was used, collected over the months of February, March and April 1995.

This sample was subdivided into four data sets as follows:

The average productivity obtained for SET 4 was, as expected, lower than the average obtained for SET 1.

In order to verify whether the average productivities of each of the four sets were significantly different, an Analysis of Variance (ANOVA) was performed with the samples. From the result of this analysis (presented in the next item) it was possible to reject the null hypothesis of equality of the four averages, that is, productivity is higher in the case of trailers and unitized loads.

The next step was to identify the relative importance of the factors on system performance. With this, it would be possible to determine which of them affects the loading productivity the most. Furthermore, it would be interesting to verify whether there is a significant interaction between the pair of factors. A simple factorial design was then carried out using statistical software (SPSS).

The factorial design allows isolating the main and secondary effects of the factors in question. The main effect indicates the importance of a given factor (type of load, for example) for the performance of the system, considering the set of possible configurations. The secondary effect indicates the possible interaction between the two factors. The results are presented in the next item.

6. RESULTS PRESENTATION

Table 1 below shows the summary of the statistics obtained after the analysis of variance (ANOVA) calculations.

Decomposing the total variation in loading productivity into the load type and truck type components, one can test whether there is a discernible difference between the four sets. In either test the extraneous influence of the other factor will be taken into account.

If the null hypothesis (H0) of equality of means is true, the ratio between the explained variance and the unexplained variance must have an F distribution.

Thus, for three degrees of freedom between groups and 664 degrees of freedom within groups, the critical value of F is 2,61 against 12,37 of the ratio between variances.

Consequently, it is concluded that, at the 5% level, there is a discernible difference between the four averages.

In the factorial design, where again the dependent variable was productivity (loading rate) and the factors were the type of cargo and the type of truck, the following results were obtained:

It can be concluded, therefore, that there are significant main effects for the two factors (significance of F equal to 0,001 for Type of Cargo and 0,002 for Type of Truck) but that there is no interaction between these factors (significance of F equal to 0,691 for the secondary effect).

Since there is no relevant interaction between the two factors, the model can be considered additive. Thus, one way of estimating the effect of each factor is to calculate the multiple linear regression of productivity as the dependent variable, as a function of the type of cargo and the type of truck. It should be noted that the two factors were treated as type 0 or 1 variables, that is, intermediate values ​​for truck size and load unitization level were not considered.

The regression results are shown in the table below:

The T statistic shows, for the two coefficients of the independent variables, values ​​greater than 1,96. Thus, the coefficients are statistically significant for a 95% confidence interval, and can be considered good estimates for the effect of factors on productivity.

The equation obtained by the regression is the following:

Productivity = 854,77*Type of Cargo + 833,56*Type of Truck + 4436,82

Finally, it can be concluded that loading unitized cargo on a given truck causes an average increase in productivity of 854,77 kg/h compared to loading fractional cargo. The effect of loading on a truck represents an average increase of 833,56 kg/h in productivity, compared to loading on a truck.

If we compare the mean variation of the samples with the global mean (M = 5326,44 Kg/h) we will have the following results, for the most favorable and least favorable configurations:

It can be seen that, if loading were carried out only on trucks with unitized load, productivity would be 15,5% higher than the current average, while if it were carried out only on trucks with fractional load, its productivity would drop by 16,1 % in relation to the average. This is another way of visualizing the influence of these factors on loading productivity.

7. CONCLUSION

As can be seen, the productivity of loading trucks at the lubricant plant is really influenced by the factors type of load and type of truck, although there may be other factors not considered in this study that also influence productivity. This conclusion can be useful, for example, in defining deliveries to be made, which can be done considering the influence of the two factors considered and the costs related to variations in loading productivity.

The results presented here show the existing potential in the implementation of a performance measurement system. Even though the influence of the load and truck factors on the productivity achieved may seem obvious to the people who work in the day-to-day handling of materials in the factory, here it could be quantified, making it possible to estimate possible gains to be made. obtained with the control of the considered factors.

Likewise, monitoring the values ​​collected for the other suggested indices, and not just the loading rate measure, may lead to the emergence of other analyzes or the perception of other influences not yet detected that could bring significant gains to the car loading sector. buckets and boxes.

A proposal for a new study would be to evaluate the influence of trucks moving between the two warehouses at the factory, in order to quantify this influence on the general loading performance. This could help decisions related to a possible change in the plant in the long term.

BIBLIOGRAPHY

* “Measuring and Improving Productivity in Warehousing” in: Measuring and Improving Productivity in Physical Distribution, National Council of Physical Distribution Management, 1984.

* Byrne, Patrick M. & Markham, William J., “Improving Quality and Productivity in the Logistics Process”, 1991.

* Caplice, Chris; Sheffi, Yossi, “A Review and Evaluation of Logistics Metrics”, The International Journal of Logistics Management, Vol.5, No.2, 1994.

* Fleury, Paulo Fernando, “Production Structure and Operational Performance: Identification of Key Variables Through Simulation”; Coppead Report 261, March, 1992.

Author Peter Wanke and Ana Luiza Mendonça