Loading...

What is Really "On-Time"? A Comparison of Due Date Performance Indicators in Production

Bachelor Thesis 2016 60 Pages

Business economics - Supply, Production, Logistics

Excerpt

Table of Contents

List of figures

List of tables

List of abbreviations

1 Introduction
1.1 Problem of the thesis
1.2 Aims of the thesis
1.3 Methodological Framework

2 Literature Review
2.1 Background and Definitions
2.2 Overview of Due Date Performance Measures
2.2.1 Output Lateness, Absolute Lateness, Squared Lateness and Tardiness
2.2.2 Relative Lateness
2.2.3 Binary Lateness and Schedule Unreliability
2.3 Classification Approach

3 Methodology
3.1 Data Overview & Data Cleaning
3.2 Lateness Computations
3.3 Interrelationship Analysis

4 Findings
4.1 Lateness Computations
4.2 Interrelationship Analysis

5 Discussion

6 Final Consideration
6.1 Results and critical reflections
6.2 Implications for further research
6.3 Implications for practice

Table of references

Appendix

List of figures

Figure 1: Lateness Variables in a Production System

Figure 2: Comparison of DDPMs

Figure 3: Relative Lateness of an Operation

Figure 4: Distribution of measured Output Lateness

Figure 5: Comparison of DDPMs (pt. 2)

Figure 6: Comparison of all DDPMs

Figure 7: Steps of Cleaning Production Feedback Data

Figure 8: Histogram: Lateness Distribution

Figure 9: Histogram: Relative Lateness Distribution

Figure 10: Scatterplots for Selected DDPMs for Company A - WS level

List of tables

Table 1: DDPM Characteristics

Table 2: Post-Cleaning Data Characteristics

Table 3: Descriptive Statistics of Company A - Work System Level

Table 4: Comparison of Mean Lateness Values

Table 5: Correlation Matrix: Company A - Work System Level

Table 6: Friedman-Test Results

Table 7: Wilcoxon signed-rank test results

Table 8: Descriptive Statistics - Company A, Order Level

Table 9: Descriptive Statistics - Company B, Order Level

Table 10: Descriptive Statistics - Company C, Order Level

Table 11: Descriptive Statistics - Company D, Order Level

Table 12: Descriptive Statistics - Company B, WS Level

Table 13: Descriptive Statistics - Company C, WS Level

Table 14: Descriptive Statistics - Company D, WS Level

Table 15: Company B - WS Actual Values Normality Test

Table 16: Correlation Matrix - Company A, Order Level

Table 17: Correlation Matrix - Company B, WS Level

Table 18: Correlation Matrix - Company B, Order Level

Table 19: Correlation Matrix - Company C, WS Level

Table 20: Correlation Matrix - Company C, Order Level

Table 21: Correlation Matrix - Company D, WS Level

Table 22: Correlation Matrix - Company D, Order Level

List of abbreviations

illustration not visible in this excerpt

1 Introduction

1.1 Problem of the thesis

In the course of the 21st century competition has risen sharply and will continue this trend. Economic complexity poses immense challenges to the majority of organizations in their pursuit of growth and profitability[1]. Therefore, outperforming competitors in the most significant competitive elements is crucial for any company.

Traditionally, quality, flexibility and costs are often seen as the deciding competitive factors and low capabilities on these dimensions most certainly lead to challenges in the striving for excellence[2]. However, in a study by Deloitte & Touche that gathered input on how management officials rank competitive factors, delivery reliability emerged as the most vital component[3].

Delivery reliability, also referred to as dependability of delivery, is defined as a company’s capability to meet the agreed on and scheduled times and quantities for delivery[4]. In other words, it measures whether delivery has been performed within a defined scope and whether target delivery dates for customers are met or not. Normally, if high safety stocks are kept, delivery reliability performance levels are strong, but additional inventory carrying costs have to be borne[5]. In Just-In-Time (JIT) production systems, inventories and stocks, as well as the respective costs are minimized, however, those systems hold a higher risk to delays and penalties for out-of-stock situations and unfulfilled customer orders. Continuously poor performance in delivery reliability leads to potential loss of customers due to low service levels, additional expenses for express deliveries and a negative impact on customers delivery reliability (downstream in the entire Supply Chain (SC)). But even early deliveries are not beneficial as they generate higher inventory costs[6]. Even though delivery reliability has such a strong impact on costs and expenses, many companies still struggle in reaching high delivery reliability levels, leaving substantial space for improvements[7].

Whether delivery due dates are met or not can be influenced by many factors. One of those is the measured due date performance (DDP), also described as schedule reliability. This indicator assesses whether orders and jobs within production processes have been executed on time and has a direct impact on a company’s performance in delivering within defined dates. This is simply due to the fact that meeting targeted delivery dates becomes more difficult, when products are made available for transport later than planned[8]. The basis of meeting due dates in production are the chosen scheduling techniques. One can define production scheduling as the attempt to use the resources available to the company in a way that best satisfies certain criteria[9]. These criteria traditionally include requirements on when production should be finished, including guidelines on late or early job completion[10]. In order to best assess the necessary throughput time, information about job characteristics and shop status are used[11]. When scheduling is done poorly, lateness in the system and with that DDP gets worse leading to little significance of measuring schedule reliability.

As the significance of DDP has been established, the question of how to measure it remains. Even when briefly reviewing existing literature on indicators of schedule reliability, one observes the vast amount of different approaches, leading to the core motivation of this thesis. The most basic measure being mentioned in a number of different articles is that of lateness, describing the difference between actual and targeted due date[12]. Baker and Christie (1984) argue that earliness should be penalized as well as tardiness and among others suggest the application of absolute lateness[13]. A third substantial approach is relative lateness. This aims at identifying the specific work system that has caused the delay and looks at the actual throughput time of an operation (output lateness – input lateness)[14]. Further measures appearing in literature are tardiness[15], schedule (un)reliability[16], mean squared lateness[17] and binary lateness (service level)[18]. Each of those indicators is defined by a specific way of assessing lateness of a production system.

Looking at the number of different methods mentioned above, various authors seem to rely on different DDP measures (DDPM) and it seems that no single approach is established in the literature[19]. Different studies use different calculation methods, leading to results that are not always comparable. This is why this thesis aims at building a first step on the way to creating a comparison of existing tools as well as an analysis of their interrelationship.

In order to improve schedule reliability, one needs to be able to define the status quo as well as measure the impact of different factors and processes on the achieved DDP[20]. Therefore, a comparison of existing measurement tools and mechanisms is necessary in order to gain insights on what ‘being on time’ in a production context really means. For theory, this is important because different studies and models on due date performance could be better comparable and combinable when the interrelationship between DDPMs is known and authors are able to use the same performance indicators. In practice, the above mentioned research draws its significance from the potential to better estimate current DDP. Once this is possible, companies could better improve their performance as they have a valid overview of their status quo[21].

1.2 Aims of the thesis

The overall aim of this thesis is to compare different measures of due date performance described in literature and used in manufacturing in order to assess their interrelationship. In respect to descriptive aims, a review of measures described in literature and a classification of those measures is to be created. The analytical aims target an in-depth analysis of lateness values computed according to different DDPMs as well as an analysis of the interrelationship between the different approaches. A discussion on the impact the choice of performance indicator has on the computed results, as well as a first guideline and recommendations on the choice of DDPM in different situations are forming the pragmatic aims.

1.3 Methodological Framework

This thesis will start off by completing an extensive literature review on DDPMs, thus deriving a funded background on different methods described in academia. A focus will be on sources from production planning and scheduling as well as production logistics. The result of the literature review will be marked by a collection and classification of the most commonly used approaches of due date performance measurements which will further be used in the second part of the thesis. With completion of this part, the descriptive aims are achieved.

The core methodology for this research project and the tool to attaining the analytical aims will be the use of empirical, statistical analyses. With the use of statistical methods the interrelationship between the different measures is to be determined. In a first step, lateness values for all selected approaches will be computed based on production feedback data from four real-world manufacturers. Further, all variables will be tested for normality using the Shapiro-Wilk test. In the following, the resulting values will be compared through statistical measures, in order to detect their interrelationships. Beginning with a Spearman correlation analysis and a Friedman-Test, an indication of the interrelationships between multiple groups (such as the chosen Due Date Performance measures) can be made. Another instrument used for testing whether central means and tendencies of two groups are coherent is the Wilcoxon signed rank test.

Ultimately, a discussion will conclude on the findings studied in prior parts of this thesis. Fulfilling the pragmatic aims, the results of the statistical analysis will be used to make recommendations on the interrelationship of the different due date indicators. Further, this thesis attempts to discuss the impact, the choice of the DDPMs has on the computed lateness values and resulting implications for research and practice.

2 Literature Review

This chapter will start off by giving an overview of the most important terms and definitions used in this thesis. Further, the state of the art in respect to DDPMs will be presented. In a last step, a classification attempt will be made, aiming at grouping the approaches found in 2.2 in accordance to their computation characteristics.

2.1 Background and Definitions

It appears that terminology in the area of production and distribution is not always conclusive when considering delivery reliability. Terms are used in different contexts and with different meaning. This is why the next paragraphs will establish the definition of key elements used in this thesis and will provide a background on the context of the research.

In section 1.1 the concepts of delivery reliability, schedule reliability, due date performance and scheduling have been mentioned. In the following each term will be explained briefly.

Delivery Reliability: Delivery reliability is denoted as the degree to which a company is capable of meeting scheduled delivery orders in terms of quantity and time[22]. It is a performance indicator concentrating on the interface to customers and becomes increasingly important as an opportunity for companies to outperform competitors and to stay competitive[23]. The concept draws its importance from its high impact on costs as for example inventory costs, express order costs or penalties on late deliveries. Due to the level of influence on a company’s success, delivery reliability is often regarded as a key logistics performance indicator[24].

Due date performance: The term due date performance has many synonyms which will be used interchangeably in the course of this thesis. Due date performance, schedule reliability and delivery capability all refer to a company’s capability to meet production deadlines and targets[25]. If due dates in manufacturing are not met and the product only becomes available at a later point, then delivery reliability is also affected as accomplishing delivery becomes more complex[26].

Due Date Performance Measure: DDPMs, also referred to as lateness measures are the indicator used to assess in what extent due date performance has been achieved in a system. Each measure represents a mathematical approach on how to quantify performance[27].

Production Scheduling: In production systems master schedules are used to note down all required outcomes, such as finished products or parts[28]. In a further step, different scheduling techniques are applied to organize production in a way that the master schedule is met. Within this framework, resource capacities and time constraints are considered[29].Without careful scheduling, production can not only be late, but some products might not be produced at all[30]. Therefore, the basic assumption of this thesis is that production scheduling has been optimized to fulfil all orders on time as long as the schedule is followed. This is due to the fact that the research aims at comparing due date performance and not scheduling quality. In order to not have the two performance elements overlap or influence each other’s values, we assume to have optimized scheduling. In this way, only one dependent variable is kept, in an environment of constants.

2.2 Overview of Due Date Performance Measures

The problem of many different ways of measuring DDP has already been broached in the first part of this thesis. However, the following paragraphs will introduce the most commonly used indicators in detail and will give an overview on their characteristics and the aspects of production they consider. Further, the concept of production lateness will be presented with a focus on different lateness variables. For good comparison, the mathematical notation of all lateness measures will be expressed by a set of fixed variables of a production throughput layout. Those variables include elements such as starting or ending points of an operation and will be introduced in the following paragraph.

To begin with, the variables used in this literature review will be introduced. Figure 1 shows an overview of an arbitrary operation (i) within production and includes planned and actual processes. The four time points on the abscissa indicate the following moments in production:

tstartplan: starting point of operation (i) as planned by the production schedule

tstart: actual starting point of operation (i)

tendplan: point of completion of operation (i) as planned by the production schedule

tend: actual point of completion of operation (i)

illustration not visible in this excerpt

Figure 1: Lateness Variables in a Production System

The indicated points in production allow for a formal description of the lateness measures described in literature. The throughput time (TTP) can be considered in two dimensions – the actual and the planned time. Each of those TTP values can be defined through two system variables, namely the time difference between starting and ending point of an operation (planned or actual, respectively).

2.2.1 Output Lateness, Absolute Lateness, Squared Lateness and Tardiness

The most basic measure is that of output lateness, also sometimes denoted as job lateness[31]. This describes the concrete time difference between the planned and actual end date of an operation (i). It is a measure often used in connection to scheduling techniques and their impact on lateness and system performance[32]. Referring back to Figure 1, one can formulate output lateness Li, out as follows:

(1)

In cases of lateness, the computed output will be negative. In the opposite case, positive values will be the result of early production completion. However, one of the main disadvantages of this DDPM is the acceptance of early production finish dates as a positive case, whilst studies have suggested the negative impact of earliness[33].

A second approach addresses the issue of the negative consequences of early completion of jobs, such as high inventory levels. Absolute lateness is calculated in a similar manner as output lateness, but as the expression suggests, it considers absolute rather than positive and negative values[34]. The mathematical notation of absolute lateness ALi of operation (i) can be expressed as follows:

(2)

As before, the point of scheduled and of actual completion are being considered. For both, early and late completion, positive values are retrieved. Absolute lateness is regarded an accuracy indicator between predicted and real values and sums up tardiness as well as earliness[35]. Multiple studies on scheduling techniques and the flow shop scheduling problem have chosen the mean absolute lateness of a production system as a performance measure[36].

Derived from absolute lateness, squared lateness proposes another possibility of looking into due date performance. However, rather than an accuracy measure this DDPM is described as a precision indicator for the level of variability of lateness[37]:

(3)

Mathematically, the squared lateness SLi of operation (i) can be seen as the squared value of either output or absolute lateness. Figure 2 shows that this indicator penalizes higher extents of lateness more, as lateness is plotted in a quadratic function[38]. Equation (3) denotes the squared lateness of an operation (i). However, approaches in literature seem to mainly rely on the mean values[39]. In order to examine the mean squared lateness, the sum of SLi for all operations is divided by the number of operations measured.

Furthermore, a large amount of literature on due date assignment rules has been considering the measure tardiness in computing the effectiveness of different approaches[40]. As the term suggests, tardiness only considers lateness of an operation (i):

(4)

For all operations completed early or on time, the value zero is assigned. In all other cases, the actual delay is counted; these are positive values. Hence, the higher average tardiness, the more delay is present in the investigated system.

Tardiness sets a focus on penalizing delays, as they can incur costs such as overtime wages, customer dissatisfaction or penalties for late shipment[41]. In concrete production systems, equation 4 can be expanded by weights for each operation’s tardiness depending on its assigned priority. Those weights can be defined by work content of an order or relations to key customers. This affects the overall mean tardiness of a system and penalizes delays in key operations or orders[42].

Tardiness is said to be one of the most standard approaches in due date performance measurement. However, with the recent trends of lean production and Just-in-time manufacturing, it has been argued that the lack of penalizing earliness marks one of its core weaknesses[43].

Figure 2: Comparison of DDPMs

illustration not visible in this excerpt

(Source: Created by the author)

Figure 2 plots the lateness measures of specific due date performance indicators for an arbitrary operation against the actual deviation in due dates (in hours). A negative value on the x-axis represents early completion of the operation while a positive value indicates lateness.

Surveying Figure 2, one can already suspect that the choice of due date performance indicator can have a strong impact on the computed lateness result. Nonetheless a statistical analysis of these tendencies will be conducted in the course of this thesis.

2.2.2 Relative Lateness

So far, all lateness measures presented take into consideration the time deviation an operation has from the production schedule in context of the entire production system. However, the total output lateness of an operation (i) comprises of two key elements[44]:

1. Input lateness
2. Throughput time deviations

The input lateness defines the delay or earliness of an incoming order in contrast to the planned starting date[45]. Subsequent operations are likely to be delayed in respect to total output lateness when they start with an input lateness[46]. Throughput time deviations mark the differences between planned and actual throughput time of an operation[47].

Figure 3: Relative Lateness of an Operation

illustration not visible in this excerpt

(Source: modified after Dombrowski 1984)

Figure 3 shows these throughput time deviations as the relative lateness between the throughput times of the planned and actual operation independent of the input and output lateness (Lin, Lout).

In the context of due date performance, these deviations are often referred to as relative lateness of operation (i) and can be described as follows[48]:

(5)

Equation 5 shows the possible dimensions of relative lateness. One can define it as the difference between actual and planned throughput time (TTPact and TTPplan), where positive values indicate lateness and negative values indicate earliness in operation completion time[49]. The second approach considers relative lateness as the difference between input and output lateness (Li, in and Li, out)[50]. Using this measure, one can assess to which extent the overall schedule situation has changed (improved or worsened) through this specific operation. This gives an indication of the individual operation contribution to the overall order lateness and can help identifying bottlenecks[51].

2.2.3 Binary Lateness and Schedule Unreliability

Another school of thought has been utilizing DDPMs which assign values not directly correlating to the extent of lateness. Concretely, a binary system is chosen when determining lateness. This means that only two possible lateness values exist where either gets chosen under specific conditions of the operation considered. The two groups of measures of this type investigated in this thesis are binary lateness and schedule reliability.

Figure 4 shows an arbitrary histogram with the distribution of operations depending on their lateness. While the blue bars represent the actual frequency of operations in that lateness range, the red line shows the cumulative frequency. This figure will be used to demonstrate the concept of binary lateness measures.

[...]


[1] (Ireland, Hitt 1999)

[2] (Lau 2002)

[3] (Deloitte 1998)

[4] (Sarmiento, Byrne, Rene Contreras, Rich 2007)

[5] (Milgate 2001)

[6] (Lödding 2012)

[7] (Lödding, Nyhuis, Schmidt, Kuyumcu 2014)

[8] (Lödding 2012)

[9] (Graves 1981)

[10] (Graves 1981)

[11] (Ragatz, Mabert 1984)

[12] (Lödding 2012; Gee, Smith 1993; Windt, Hütt 2011)

[13] (Baker, Christy 1984; Gee, Smith 1993; Liao, Lin 1998)

[14] (Dombrowski 1988; Lödding, Nyhuis, Schmidt, Kuyumcu 2014)

[15] (Lödding 2012; Ribas, Companys, Tort-Martorell 2013; Vig, Dooley 1991)

[16] (Lödding 2012)

[17] (Gee, Smith 1993)

[18] (Jodlbauer, Huber 2008)

[19] (Baker 1984)

[20] (Hon 2005)

[21] (Lödding, Nyhuis, Schmidt, Kuyumcu 2014)

[22] (Handfield, Pannesi 1992)

[23] (Nyhuis, Wiendahl 2008; Hon 2005)

[24] (Lödding 2012)

[25] (Lödding 2012)

[26] (Schmidt, Bertsch, Nyhuis 2014; Lödding 2012)

[27] (Baker 1984)

[28] (Galloway, Rowbotham, Azhashemi 2012)

[29] (Graves 1981)

[30] (Günther, van Beek 2003)

[31] (i.a. Lödding 2012; Gee, Smith 1993; Münzberg, Schmidt, Beck, Nyhuis 2012)

[32] (Dileepan, Sen 1991)

[33] (Graves 1981, Lödding 2012)

[34] (i.a. Liao, Lin 1998; Gee, Smith 1993; Sha, Liu 2005)

[35] (Sha, Liu 2005)

[36] (Botta-Genoulaz 2000; Sha, Liu 2005; Kellegöz, Toklu, Wilson 2010)

[37] (Sha, Liu 2005)

[38] (Song, Hicks, Earl 2002)

[39] (Sha, Liu 2005; Gee, Smith 1993; Song, Hicks, Earl 2002)

[40] (Ribas, Companys, Tort-Martorell 2013; Vig, Dooley 1991; Baker, Scudder 1990; Mattfeld, Bierwirth 2004)

[41] (Vig, Dooley 1991)

[42] (Mattfeld, Bierwirth 2004)

[43] (Baker, Scudder 1990)

[44] (Nyhuis, Wiendahl 2008)

[45] (Lödding 2012, Bertsch, Schmidt, Nyhuis 2014)

[46] (Nyhuis, Wiendahl 2008)

[47] (Lödding 2012)

[48] (Dombrowski 1988)

[49] (Nyhuis, Wiendahl 2008; Wiendahl, Ludwig, Ullmann 1994)

[50] (Bertsch, Schmidt, Nyhuis 2014)

[51] (Scholz-Reiter, Windt, Liu 2011)

Details

Pages
60
Year
2016
ISBN (eBook)
9783668479098
ISBN (Book)
9783668479104
File size
1.2 MB
Language
English
Catalog Number
v370576
Institution / College
Jacobs University Bremen gGmbH – Production & Logistics Networks
Grade
1,0
Tags
Production Due Date Performance Reliability

Author

Share

Previous

Title: What is Really "On-Time"?  A Comparison of Due Date Performance Indicators in Production