Integrated classification methods for spare parts

(1)

Integrated classification methods for spare parts

A case study on a mass production factory

Selin Yesilkayali

Document type: Independent degree project – second cycle

Main area: Department of Information Systems and Technology (IST) Credits: 300 ECTS

Semester, year: 10, 2020 Supervisor: Aron Larsson Examiner: Leif Olsson

Degree program: Master of Science in Engineering; Industrial Engineering and Management

(2)

ii

Abstract

Inventory management is a complex system which involves different stakeholders from multiple areas in a company which creates a limitation when seeking information between involved staff. Having the right procedure of tracking regular and critical spare parts will give a better control and efficiency in the production process. It is important to have the right classification method to facilitate critical spare parts. The incorrect criteria classification can be achieved in case inventory management have the wrong systematic procedure.

Classification methods have different purposes and achieve the highest utilization by combining a variety of methods. By integrating classification methods, set limits and combination of multiple criteria decision analysis can be performed. The study has conducted a case study to compare and evaluate the performance of inventory management in a trustworthy and efficient way. A theoretical framework is constructed with the intention on identify which classification methods can be combined and applied to a production factors criterion. Based on interviews with stakeholders from maintenance, warehouse, and production area related to spare parts and the company’s software system.

Two perspectives were used to map the qualitative and quantitative measures.

The results show 14 criteria were defined as parameters that measure the performance of criticality in spare parts. The conclusion of both perspectives suggests combining and implement an integration of AHP and ABC classification methods. A proof of concept is demonstrated on AHP analysis and ABC analysis to identify the critical spare parts and the criteria.

Keywords: Inventory management, maintenance, warehouse, production, AHP analysis, ABC analysis, VED analysis, criteria, criteria analysis, spare parts.

(3)

iii

Sammanfattning

Lagerhantering är ett komplext system som involverar olika intressenter från flera områden i ett företag. Detta skapar en begränsning av information som de involverade personalen har möjlighet att nå. Att ha rätt arbetsmetod för att kunna spåra vanliga och kritiska reservdelar ger möjligheten till en bättre kontrollstyrning och effektivitet i produktionsprocessen. Det är viktigt att ha rätt klassificeringsmetod för att kunna underlätta identifiering av kritiska reservdelar. Vid användning av fel klassificeringsmetoder kan kriterier bedömas fel. Klassificeringsmetod har olika syften och uppnår det högsta utnyttjandet genom att kombinera olika metoder. Denna fallstudie syftar till ett nytt tillvägagångssätt som utvärderar vilka metoder som kan implementeras för att bidra till ett mer pålitlig och effektiv lagerhanteringen. Ramverket för denna studie är avsett att identifiera vilka klassificeringsmetoder som kan kombineras och tillämpas på företagets kriterier. Denna metod involverar intressenter från underhåll, lager och produktionsområdet relaterat till reservdelar. Ett teoretiskt ramverk konstruerades för att identifiera de kriterier som krävs för tillämpning av klassificeringsmetoderna med flera kriterier. Genom att analysera kvalitativa data kunde intressenternas kriterier identifieras. Dessa bestod av 14 kriterier som används inom reservdelar. Två perspektiv av ramverk användes för att kartlägga de kvalitativa och kvantitativa data. Resultaten som visas från båda ramverken föreslår att fallföretaget bör kombinera och implementera multi- kriterier metoderna AHP och ABC. Ett bevis på konceptet visas genom att demonstrera AHP analys och ABC analys för att identifiera de kritiska reservdelarna och kriterierna.

Nyckelord: Lagerstyrning, underhåll, lager, produktion, AHP analys, ABC analys, VED analys, kriterier, kriterieanalys, reservdelar.

(4)

iv

Acknowledgement

I would like to thank my supervisor Aron Larsson at Mid Sweden University for the time and advice he has contributed through the study. I would also like to thank the case company and all the participants that were involved in this research study.

(5)

v

List of abbreviations

ABC Activity Based Costing AHP Analytic Hierarchy Process VED Vital Essential Desirable

(8)

1

1 Introduction

Following chapter is presenting background a practical background about the research area and research problem, followed with aim of the study.

1.1 Research background

Spare parts are a significant part of a company’s inventory and crucial for production; they require constant control and development (Roda et.al., 2012;

Stanford & Martin, 2007). When a certain component fails, the need for spare parts arises then a decision gets made as to whether a machine requires replacing or repairing. A spare part is a unit whose purpose is to replace a damaged part of a machine. Maintenance staff assists to restore the system which will then continue to perform its intended function (Roda et.al. 2012; Roda et.al. 2014).

Machine availability and performance will be at optimum levels when they are maintained properly. This is an important task to make sure there is continuity of production. By correctly tracking the availability of spare parts, machine supervision will be at an optimum level (Antosz & Ratnayake, 2016).

The problem within inventory management is usually deciding how the inventory should be managed. A useless inventory of spare parts can have adverse financial consequence (Grondys, 2015). These are consequences that most companies are usually not aware of, nor do they have the knowledge or the competence to control the inventory effectively (Mikaelsen, 2015).

One of the many reasons for failing to achieve an optimum inventory of spare parts is that managers do not know how to maintain an inventory. Companies usually manage all spare parts the same way which is a mistake that can be corrected over time by developing procedure on evaluate the character of spare parts (Mikaelsen, 2015; Grondys, 2015).

1.2 Research problem

Spare parts management can be a complicated task, some problems can arise when there is uncertainty about the demand for spare parts when there is large inventory list that causes difficulty in predicting what will be needed (Roda et. l. 2012; Roda et al. 2014).

Industries which aim for continuous mass production can face potential loss of profits during unavailability of spare parts. A longer wait of spare part can cease to a long unproductive downtime and stop on machines (Roda et. l. 2012; Roda et al.

2014; Hu et al. 2017). Mass production industries producing goods nonstop push the limits of their machines to meet the demand. At the same time, the demand for maintenance increases in order to restore machines to their original condition and to continue operating on the required performance (Lopes et al. 2019).

(9)

2 Research find general solutions for keeping units in stock instead of focusing on spare parts (Hu et al. 2017). With the right identification and proper method of spare parts classification the importance of spare parts can be conveyed effectively (Antosz & Ratnayake, 2016). To minimize unavailability, and to improve performance of machines with high utilization, a critical evaluation and ranking of equipment can be performed. This way, those doing the maintenance will have a better understanding of the most critical spare parts. Having the wrong classification method can cause less prioritization of spare parts, increase mechanical failures, and have a negative impact on production. This means an appropriate classification assessment needs to be developed, used correctly, and updated regularly to ensure good maintenance management (Lopes et al. 2019).

Industries are still struggling with integrated information systems in classifying spare parts efficiently (Roda et al. 2014).

With there being a lack of research with the focus classification methods and their properties this study will contribute to identify which integrated classification methods and featured properties can be used for spare parts.

1.3 Aim and Research question

This study will investigate and find recommendations for mass production industries on how integrated classification methods can be implemented to classify crucial spare parts.

• What systematic classification methods for spare parts are more feasible for a mass-production company?

1.4 Delimitations of the study

This study will be limited to a case study with the focus on mass production industry where production needs to be continuous and in any extreme cases where critical consequences might arise if production stops. Furthermore, the methodology and conclusions in this study will be generally applicable in similar industries.

(10)

3

2 Theory

Following chapter will present the theory of this study and delve into deeper knowledge surrounding this case study.

2.1 Spare parts management

Spare parts are potential replacements for working parts in machines that get damaged. According to Roda et al. (2012), the management of spare parts can be maintained as follow.

• The identification of spare parts can be done by part coding and classification.

• Identifying and developing the stock management system and managing and formulating inventory control policies and systems.

• Implementing a computerized maintenance management system that can identify and manage spare parts.

• Implementing a processing and inventory control systems operation that manages information and policy testing which will continuously be improved according to its performance. The system determines when to place an order and how many parts will be needed.

2.1.1 Spare parts characteristics

The maintenance staff is responsible for keeping equipment in operating condition - if staff fails to do this there will be a delay in production. There are certain reasons for the lack of controlling spare parts that exist within the maintenance. Sometimes it is uncertainty in understanding when a part is required and sometimes in which quantity it is needed. Usually, the maintenance has a too large inventory list of spare parts which makes controlling and upkeeping a tedious job. A limited number of suppliers for a spare part can cause the sourcing to be much harder resulting in loss of lead time and cost. There is also a problem with having multiple suppliers since the quality of the spare parts can variate which can affect the lifespan of the spare part and in some circumstances increase costs. It is difficult to determine the stock unit of spare parts because of the obsolescence machine (Roda et al. 2012; Roda et al. 2014).

Wrong observations and characterization are normally determined. The company can moreover have a lack of poor inventory data recordings, information visibility by storing data in separated software, or bad description of how to manage the process of tracking inventory (Roda et al. 2012; Roda et al. 2014).

(11)

4

2.2 Spare parts classification methods

Molenaers et al. (2012) explain that industrial plants are late with implementing new classification methods and they are still using old classical methods.

Classification is a step to bring attention on spare parts which are important and enables decision-making process. Classification method allows to control spare parts differently by making a criticality analysis. The term criticality has different definition depending on how it is undertaken (Roda et al. 2014). Teixeira et al.

(2017) and Roda et al. (2014), mention that there does not exist any systematic or well-structured procedure for evaluating criticality of spare parts. A lack of systematic procedure can create a wrong criteria classification. In an industrial plant, it is important to have a proper classification which can categorize and organize spare parts. Having a useful tool that performs the mentioned tasks will give better control of inventory and pay more attention to the critical spare parts (Stoll et al. 2015).

There are several methods to be used when categorizing spare parts (Stoll et al.

2015). One of the methods are divide spare parts into process criticality and control criticality. Process criticality is described to be anything related to production loss and damage to the environment. Meanwhile, control criticality relates to the immediate availability of spare parts and this part is difficult to control in case a critical situation occurs (Teixeira et al. 2017; Molenaers, 2012).

Another categorization method is dividing the spare parts into three category which are “high”, “medium” and “low”. This method considers the number of machines that have a spare part installed with a critical criterion. A spare part can be counted as critical when it is installed on multiple machines, when a spare is unavailable it will affect all the machines (Stoll et al. 2015).

It is a crucial task to assort spare parts into relevant categories to control a wide and highly diverse inventory. Spare parts as well as their assortment can differ based upon their criteria from maintenance perspective and warehouse viewpoint.

Maintenance focus on parts that can have severe consequences for the company and production in case any unavailability exists during their decision-making process meanwhile the warehouse focus on assortment and control criteria of the spare parts. (Molenaers, 2012; Roda et al. 2014).

2.2.1 ABC analysis

Activity based cost (ABC) analysis is a classification method known for its traditional used in certain industries. ABC helps to simplify stock management and allows for strict control of critical spare parts. It is further known for being a one- dimensional criterion. This means the result might not be useful in practice (Teixeira et al. 2017).

(12)

5 ABC analysis is based on the Pareto principle which classifies inventory into A, B, and C depending on item value and consumption value. The grouping of classes is usually based on an 80/20 percentage rule, there does not exist any prevalence on which specific percentage should be used in the classes of A, B, and C (Stanford &

Martin, 2007).

This method helps to group the inventory according to criticality and the items with the highest value, see Figure 1 (Ayat, 2017). The consumption value, which is associated with the higher class, will need to receive more attention than the rest of the classes (Stanford & Martin, 2007).

Items in group A are the inventory with the highest consumption value with 80%

and inventory costs value of 20%. The inventory in class A receives the most attention regardless of the other classes because items in this class have the highest consumption value and need to be handled carefully. Class B includes 30% of the consumption value with 15% of inventory costs value with moderate importance.

Items in class C include about 50% of consumption value with 5% inventory cost value with low importance. Class B and C do not need the same attention as class A, as those two classes can be monitored with a low safety stock policy (GunGöner

& Dagdeviren, 2017).

Figure 1 - Graphical illustration of ABC-analysis (Ayat, 2017).

2.2.2 VED analysis

Vital, essential, and desirable (VED), analysis is a commonly used tool in maintenance. This method is known for being sorted practically by categorizing and scoring which are done with the help of maintenance expert consultation (Teixeira, 2017; Gajpal et al. 1994). Subjective judgment is used during expert consultation, which can be problematic when all the experts has different viewpoint on judgement (Teixeira, 2017).

There are three different critical classes: present Vital, Essential, and Desirable (Ayat, 2017). “Vital” includes spare parts that have a more significant impact on the

(13)

6 production process. Spare parts stock-outs in this class can result in substantial losses (Teixeira et al. 2017).

“Essential” includes spare parts with lesser importance to the production process.

Stock-outs on spare parts do not affect these as much but can cause a moderate loss (Teixeira et al. 2017). “Desirable” is the class with the least disruption and risk during the production process (Teixeira et al. 2017). Machines can still be operable without those two classes of spare parts; but in the long run, there could be serious problems with the machine’s operational capabilities (Gajpal et al. 1994).

Table 1 presents an example from Teixeira et al. (2017) on the structure of VED analysis. The example is a list of spare parts being evaluated according to criteria such as “function” and “impact on production”. It is then scored according to criticality. “Function” criticality has a score from 1 to 3, and “production impact”

has a score from 0 to 3. By taking the sum of each row, a grouping can be made relative to VED analysis. All scores under 2 are “desirable,” scoring between 3 and 5 is “essential,” and the highest score is “vital.”.

Table 1 – List of spare parts with VED-analysis (Teixeira et al. 2017).

2.2.3 AHP analysis

Analytic hierarchy process (AHP) analysis is a decision-making tool “used in a wide range of fields, mainly in operations management. It solves complex decisions by prioritizing alternatives” (Teixeira et al. 2017). The method is based on expert judgments that help the decision-maker to determine which decision alternatives give the best solution to the problem and reach company goals (Teixeira et al. 2017).

It is designed as a hierarchy diagram with a goal on the top and consisting of such levels as criteria, sub-criteria, and alternatives (Nurcahyo et. al. 2018).

(14)

7 In Figure 2 there is an illustration of a hierarchical diagram about choosing the best car. On the first level, the criteria are defined with a goal or requirements that must be fulfilled. In the second level all the automotive alternatives are set out (Andersson, 2016). A sub-criterion can be added in between criteria and alternatives. The purpose of having sub-criteria is to expand preferences. For example, color criteria compare the importance of different colors or price criteria compare to the price of purchasing a new or old car (Andersson, 2016).

Figure 2 - Diagram of analytic hierarchy process (Andersson, 2016).

This tool has a basis for determining relative priorities or weights between the criteria sharing the same predecessor in the hierarchy. Stakeholders will do a pair- wised comparison between different criteria and determine the strength of the criteria with help of Table 2 (Erdinc, 1996; Molenaers et al. 2012; Partovi & Burton, 1992).

Table 2 – Pairwise comparison scale.

A comparison matrix will then be performed to measuring each of the criteria relative important toward decision making (Gajpal et al. 1994). Table 3 explains how the comparison between the weights is being calculated. The results of the weighting will then be summed up in each column to later be normalized and to calculate the eigenvector, also called a priority vector, of each criterion.

Table 3 - Pairwise comparison between weights of criteria (Andersson, 2016).

(15)

8 Normalization is calculated by equation 1, where the total summation of the criteria column is divided with their origin weight, with the same calculation procedure being done on all the criteria. The priority vector is then calculated by the mean of the whole row for each criterion. Equation 1 explains that A is dimensional n of the comparison matrix, w is the priority vector to eigenvalue λmax (Erdinc, 1996;

Nurcahyo et. al. 2018).

(1) Further, the consistency index and consistency ratio are then calculated. The purpose of calculating the consistency index ratio is to understand how consistent the decision-maker or stakeholder is when the decision is being made. The consistency index is calculated by equation 2 where λ^maxis the highest eigenvalue and n is the number of criteria. Consistency ratio measures the coefficient degree of homogeneity among the judgments issued from the weights. This can be calculated through equation 3 where the consistency ratio is divided by the random index from Table 4 (Erdinc, 1996; Nurcahyo et. al. 2018). AHP decision-making tool will be furthermore explained in upcoming chapters.

(2)

(3)

Table 4 – Random index scale (Erdinc, 1996).

2.3 Integrated classification methods

Spare parts can be classified with different methods to enable proper control without excessive work (Hagberg & Henriksson, 2010). Classification methods have different purposes and combining a variety of methods can give a better result (Hagberg & Henriksson, 2010). The principle of using multiple-criteria decision analysis is to integrate different classification methods to create a limitation in the selection of criteria (Stoll et al. 2015; Roda et al. 2014).

Figure 3 is presenting the process of how to perform multiple-criteria decision methods. The first step is to understand the necessity and importance of the spare parts for maintenance. Then identify the criteria based on maintenance and inventory management opinions to classify and control the criticality of spare parts.

Classification methods will be integrated and group the spare parts that are sharing the same stock management policy and with results give better control on critical spare parts in the companies’ inventory management (Teixeira et al. 2017).

(16)

9 Figure 3 - Forming of multiple-criteria classification method (Teixeira et al. 2017).

2.3.1 ABC analysis combined with VED analysis

A matrix will be performed when combining ABC analysis and VED analysis. The purpose of this method is to give appropriate control over and supervision of spare parts. The combination will be performed in three stages. In first stage the VED analysis is performed by defining criticality of the spare parts by the expert judgment. In the second stage the ABC analysis is calculated based on the inventory criteria. Then both results are combined in one cross-joint to understand the ABC- VED matrix analysis (GunerGönger & Dagdeviren, 2017).

The first category in the ABC-VED matrix is presenting items that are expensive and vital, they are named AV, AE, AD, BV, and CV. The second category will present the essential items that have an average value, which are named BE, CE, and BD. The last category will only have one sub-category which is CD, see Figure 4 (GunerGönger & Dagdeviren, 2017).

Figure 4 - Combination of ABC and VED classification (GunerGönger &

Dagdeviren, 2017).

2.3.2 AHP analysis combined with VED Analysis

An absolute performance will be calculated when using AHP analysis and VED analysis. AHP analysis will be calculating the expert judgment by pair-wise comparison and weighting the different attributes. The result will give the decision- maker the opportunity to determine which criteria are the most important alternative. The method is designed in a hierarchical diagram with a goal on the top level, while the first level below the goal statement is the criteria. A second level is created in case a sub-criterion is included in the research. The third level will contain a calculation of VED analysis (Nurcahyo et. al. 2018).

In the first level, a pairwise comparison is calculated in a matrix form between the criteria. Pairwise comparison is arranged and compared through stakeholders or decision-makers qualitative judgment. By calculating the matrix, a priority vector will be given as a solution (Nurcahyo et. al. 2018).

(17)

10 In case of a sub-criterion existing, the priority vector with the highest score will be calculated. A second pairwise comparison is made with the sub-criteria that will result in a new priority vector. The third level is computing VED analysis, see Figure 5. Depending on which priority vector had the highest score in second level, the stakeholder will be classifying spare parts according to Vital, Essential, and Desired categories. Then the last pairwise-comparison is made between the three categories. The Vital category with the highest appraisal will be the critical ones (Nurcahyo et. al. 2018).

Figure 5 - Combination of AHP and VED classification (Nurcahyo et. al. 2018).

2.3.3 AHP analysis combined with ABC analysis

A standard AHP calculation will be carried out to rank and determine the weights of the criteria. When the calculation is done an ABC analysis will be performed with the priority vectors that have been calculated using AHP method (Partovi &

Burton, 1992).

As in chapter 2.2.3. a hierarchy diagram is designated that contains up to three levels. The first level calculates the pairwise comparison between the criteria chosen to achieve the goal. Stakeholders or decision-makers are scored by qualitative judgment. All the measurements will then be mapped in a matrix model and by calculating the matrix model, the priority vector will then be obtained. In the second level with sub-criteria, another pairwise comparison will be calculated using the matrix model and the criteria with the highest priority vector (Erdinc, 1996; Partovi & Burton, 1992; Flores et al. 1992).

In the final level, the criterion value will be multiplied with their corresponding priority vectors that were calculated in AHP method. Spare parts data with information for each criterion is then collected. Then it is calculated by a single criteria value x divided by the total in criteria value x multiplied by the corresponding priority vector, see equation 4. This calculation is made on each i^th spare part. Lastly, a total summation f of each criterion p for one spare part is calculated which can be seen in equation 5. The total calculation will then be placed in descending order to define the A, B, and C classes (Erdinc, 1996; Partovi &

Burton, 1992; Flores et al. 1992).

(18)

11 (4)

(5)

In the example below, Figure 6 shows how to determine the best ABC analysis of stock-keeping units for management. A qualitative and quantitative criterion is chosen. The final score obtained from the AHP analysis will be classified according to the A, B, and C categorizations (Partovi & Burton, 1992).

Figure 6 - Multiple-criteria decision method using ABC and AHP analysis (Partovi & Burton, 1992).

2.4 Quantitative and Qualitative criteria methods

Inventory control plays a significant role in operations and manufacturing; two areas companies often have problematic issued with. The inventory covers anything related to purchasing of new items or, spare parts and anything needed for the production process. Companies’ manufacturing activities will benefit from having better inventory control (Wild, 2017).

Each classification method has support properties (i.e., what assessment criteria they intend to operate on) and the classifying method can be based on quantitative or qualitative criteria. For instance, a single classifying property cannot define the entire criticality of a spare part (Teixeira et al. 2017). Roda et al. (2014) explained that spare parts classification cannot be based on a single property. Integrated classification methods which can perform multiple criteria can manage well- structured classification of spare parts.

Roda et.al. (2014) investigated eighteen works of literature and concluded that each classification method has different properties. Lopez. et al. (2019) has done a similar investigation on AHP analysis with different literature reviews, which proposed

(19)

12 the method using different properties. The authors highlighted that they cannot find any relation or pattern among the chosen property to fulfill the main objective with the help of the criticality assessment.

2.4.1 Qualitative criteria method

Qualitative criteria are determined from a maintenance viewpoint (Roda et al.

2012). Cavalieri et al. (2008) asserted that there are few assets to be considered during a qualitative aspect. This method attempts to prove the importance of keeping spare parts in stock. The authors mentioned further that it is important to understand factors that are influence management such as on cost, downtime, or storage considerations.

Classification methods such as VED and AHP analysis are qualitative criteria methods identified through maintenance perspectives and experiences viewpoints (Roda et al. 2012). The criterion for VED analysis can be identified through criticality or loss of production (Madan & Ranganath, 2014). Application of VED analysis task of controlling spare parts management and manufacturing equipment (Madan & Ranganath, 2014). For literature analysis, the common properties can be used out for the classification methods, see Table 5 to 8.

Table 5 - Classification criteria support properties for AHP, (Praveen et al. 2016).

Table 6 – Classification criteria support properties for AHP, (Perez et al. 2020).

Table 7 – Classification criteria support properties for AHP, (Hu et al. 2020).

(20)

13 Table 8 – Classification criteria support properties VED, (Praveen et al. 2016).

2.4.2 Quantitative criteria method

Cavalieri et al. (2008) asserted that quantitative criteria are determined through technical variables. This method is preferred when single criteria is being used such as annual demand, annual purchasing cost, or obsolescence of spare parts.

ABC analysis is an example of a quantitative criteria method that has a single criterion. The properties can be anything related to inventory management such as unit price, annual demand, annual purchasing, or demand volume (Roda et al.

2012).

Madan & Raganath (2014) asserted that the properties for ABC analysis can be identified through anything related to the production processes as purchasing, selling, or costs. The levels of spare part criticality can be identified by annual maintenance budget. (Cavalieri et al. 2008). Table 9 and Table 10 are featured properties according to the classification method.

Table 9 - Classification criteria support properties for ABC (Roda et al. 2014.).

Table 10 - Classification criteria support properties for ABC (Partovi & Burton, 1993.).

2.5 Literature review

Ayat (2017) explained that managing spare parts inventory is a complex area, since it contains varied numbers of spares with low demand. Further, it is important to carry an inventory of spare parts, since a situation where spare parts are out of stock can cause disruption in production and financial loss for the industry. The author has analyzed the inventory management of a printing company with the aim of implementing stricter management control by minimizing dead stock and increasing the optimal availability of items. The paper used a combination of ABC and VED analysis in a matrix form. They begin by calculating the annual values in

(21)

14 ABC analysis, and then they classify the inventory with help of VED analysis where they categorize items depending on their stock out cost, nature of the product, source of supply, and lead time. Lastly, it is framed in a cross-tabulating matrix form where they combine both techniques. Data has been collected through interviews with the maintenance staff, production area, purchasing department, supervisors, and workers. Based on results from the analysis an elimination of dead stock, minimization out-of-stock situations, efficient decision making in purchasing, and reduced capital investment has been conducted.

Teixeira, Lopez, and Figueiredo (2017) showed that inventory management and management of spare parts can be a complex area of study. The authors explained that there are difficulties in collecting data, since a large number of spare parts are involved. It is important to understand factors that influence problems of production loss, quality loss, and costly inventory levels in order to eliminate those factors. The paper used a combination of VED and AHP analysis, starting with defining production criteria by dividing it in two parts and defining different levels: one criterion for function and one for impact. The results form a matrix where VED analysis is used to classify spare parts. After validation of criteria, a comparison is made with the help of AHP analysis, which used criticality, lead time, and price as subcriteria to choose the most appropriate stock management policies. Data has been collected through maintenance. The study has resulted in helping the organization decide on the basis of quantitative information how they can make decisions that will keep spare parts in stock.

Shashikumar, Sarkar and Sanyal (2017) explained that mass production industries are facing global competition, which leads to increased productivity at reduced cost. The paper has been analyzing facilities layout in a manufacturing industry to tie in investment by using multicriteria classification methods combining ABC analysis with AHP analysis. They begin with using AHP analysis to understand the relationship between goals, activities, and costs to create the integration of multicriteria classification methods. Later they use ABC analysis to understand the relationship between costs and activities. Lastly, they combine both methods to get a more specific result. The study has been formed around the management goal of improving capacity, quality, productivity, and the flexibility to eliminate any production stops or bottlenecks since it considers a mass production industry. They conclude that management can make better decisions by integrating activities and strategic goals into investment analysis.

2.6 Case Company

The case company’s process of fulfilling the need for spare parts arise when the operator detects machine failure or production delay, see Figure 7. Then the operator creates a faults report in the maintenance system, and the maintenance supervisor reads it to conclude whether it is an emergency work or planned work.

(22)

15 Then supervisor informs maintenance staff to analyze further it is a mechanical or electrical fault in the machine. It is important for maintenance staff to have the right information so as to not cause delays by choosing the wrong spare parts.

When they have detected the fault, the supervisor will give the faults report to the production planner. In a planned work, the production planner will do research on which spare parts are needed and to see if they are registered in the business system. By checking the business system, he will know if there are any spare parts left in the warehouse. Then he will then create a pickup order on the spare parts and send an email to the warehouse. The warehouse then saves the spare part for the specific workorder. The planned work will then be scheduled and sent to the responsible supervisor and department leader.

It is mandatory to have a workorder when repairing the machine and taking spare parts from the warehouse. The maintenance manpower will receive the workorder from their supervisor. Maintenance will then take the spare parts from the warehouse and use them to restore the machine to working order.

Figure 7 - Process flowchart on repairing machine.

If the wrong spare part is chosen, maintenance will replace it with the right unit, but if the correct unit is out of stock, they will devise a temporary solution and tell the warehouse to order a new one. If a temporary solution is not possible, then the machine’s restoration will be postponed until the spare part is delivered.

Minimum stock and order quantity of spare parts are decided on with the help of information provided by the supplier of the machine or by the maintenance department. The supplier can also provide information about alternative suppliers in case a particular spare part is out of stock. To prevent a stockout, the warehouse creates new orders each time a spare part hits its minimum stock. The business system informs the warehouse automatically when they hit minimum stock level.

All the information about the supplier, minimum stock, and the price is in the system.

(23)

16

3 Method

Following chapter will presenting the methodology of this study and order to fulfill the research question.

3.1 Research strategy

There is various research strategy with different purposes which can be used in a study. One of the strategies are case study which has the approach of study a specific case. Empirical research is carried out within the research field include of industries, municipalities, and other governmental organization (Denscombe, 2017).

A case study will be upheld in this study with the purpose of gain a deeper insight on challenges the industry factor faces. The aim is to focus on detail the production area whereas’ being business critical for the company. This research strategy considered to be most suitable as it gives the flexibility to use different methods to collect information. Other strategies will require to in the field sights to observe and collect data (Denscombe, 2017).

3.2 Data collection method

Information will be collected based on answering the research question and procedure. Denscombe (2010) explained that use of multiple method can strengthen the research by having information from different perspectives. A convergent parallel mixed method will be used in this study which entails the researcher to collect qualitative and quantitative elements during the same research process.

The approach of a convergent parallel mixed method is to collect both qualitative and quantitative data and analyze these separately to find out if the findings can be confirmed or disconfirmed on each other. Qualitative data can be any form of data such as interviews, observation, or records. Quantitative data includes instrument data, observational checklist, or numeric records as census data, see Figure 8 for illustration of the steps (Creswell, 2014).

The purpose is to collect both forms of databases by using same or parallel variables. Both methods use the same concepts during the data collection process.

Final procedure can be done by merging the two forms of collected data into one table or graph. This method is called a joint display of data, where the idea is to jointly display both forms of data in an effectively by merging them in a single visual (Creswell, 2014).

(24)

17 Figure 8 - Illustration of convergent parallel mixed method (Creswell, 2014).

Interviews has the purpose of obtain detailed material and can be performed either as structured, semi-structures or unstructured. Semi-structured interviews give the opportunity for the researcher to create a predetermined question but with a flexibility to add question during or after interview (Denscombe, 2017). Using semi- structured interviews will allow the researcher during this study to gain valuable insight on the inventory management and criterion the case company is living upon when they classify criticality of machines and spare parts: how the company identify spare parts, and which are the most crucial machines and units.

Interview questions will be formed through inspiration from previous research within same subject and documents from the case company which are written internal on how the process is carried throughout the departments. A better research insight will be given by having access to information about the documentations from the case company. Further information will be collected from stakeholders with necessary experience within the production and spare parts areas, concurrently information from the company’s software system will be collected at the same time.

The participants for the interviews will be selected based on their knowledge and experience to provide valuable information relevant to research question. To gain a holistic view participant from different roles experienced within inventory and spare parts management will be selected. If necessary, questions will be reformulated to the participants role.

Numeric records will be collected during the research to test and observe if the results from research confirm or disconfirm. The purpose is to reflect and gain holistic explanation on attitude and decision-making. Concurrently information from the company’s software system, connected to a database, will be collected during the research.

All interviews will be transcribed into a text document to remain data order to perform a data analysis. Transcription means that audio recordings are converted into text format analyze the interview material. It is a valuable use since it enables

(25)

18 the researcher to find new information in an easier process and the opportunity to pay attention to what the respondents say and how they express (Denscombe, 2010;

Bryman, 2011). All numeric record will be saved in Excel-documents to obtain an easier analyzation of data.

3.3 Data collection procedure

A literature analysis will be done to understand which feasible classification methods can be used for this case study, with a further analysis the qualitative and quantitative criteria method can be identified. By understanding the purpose of each classification method and their properties a delimitation can be done related to mass production factory.

This study will only use one quantitative criteria method and two qualitative criteria methods, see Figure 9. The reason for using two qualitative criteria methods is because the results from qualitative interviews can differ from each answer. A closer study will be done on the field of inventory management to understand which properties each method is using and then portray them in a framework. This will help to understand how to categorize the criterion and classification method when portraying the criteria chosen from interviews.

Figure 9 - Method for spare parts analysis.

Interview questions will be tested on maintenance engineer with the knowledge of spare parts in order to judge whether the questions are relevant, understandable, or need to be changed or added to. Interview with the right stakeholder with relevant experience and knowledge about machines and spare parts will be uphold, see pre-determined questions in Appendix A to D. Participants of interviews will be with six stakeholders consist of property owner, maintenance manager, maintenance engineer, maintenance production manager, production planner and warehouse manager. Before the interviews, stakeholders will be informed the purpose of study and ensure they understand their rights according to ethical aspects. All the participants will have the right to discontinue or disapprove questions.

The interviews will be conducted in two parts; the first part will be an individual interview with each participant, and the second part will be an interview with all

(26)

19 the participants together. The interviewer will have the opportunity to ask additional questions depending on the participants’ answer. All interviews will be held face-to-face and in Swedish to eliminate misunderstandings, further each interview will be recorded with approval from participants and transcribed for the opportunity to analyze the documentations and answers (Bryman, 2011).

All data will be collected and analyzed to understand the context of the company and the production goals according to set criteria for each objective. A second discussion will be set up with the participating stakeholders to get a final approval of the findings. This is called respondent validation and is a confirmation that the results and impressions on researchers’ part are consistent with the participants answers (Bryman, 2011).

Participants will be ranking the machines depending on their criticality for the production. Since the participants will have different views, ranking will be calculated and the most common machine on the top three will be investigated. If necessarily, further analysis of the interviews will be done to narrow the focus down to one machine that has the most impact to the production. Critical spare parts will be chosen with help of property owner and since one machine has hundreds of units the focus will remain on the ten relevant ones with the major dependency on different machines and related to production.

An inspection of historical data banks will be done together with maintenance manager to eliminate collection of wrong numeric records. Data will be collected through the maintenance, business and stoptime system. It will give an understanding of which machine had the most faults report during 2019. By focusing on the machine with the frequent stops a better monitoring can be done before it affects the entire production line. A machine which stops frequently require more attention, unnecessary man-hours and work which affects the whole production and create bottlenecks (Dynamox, n.d.). A closer analysis will be done to find out which machine units had the most breakdown.

The collected data will be displayed in the proposed framework and the results validated to understand which integrated classification each machine will have.

When all the data is displayed in the relevant frameworks, a better understanding will be available on which method is suitable for each machine. A clear comparison can be made with the interview data and system data. See Figure 10 for an illustration of the data collection procedure.

(27)

20 Figure 10 – Data collection procedure of research study.

3.3.1 Data classification procedure

The application of the research method consists of two parts. The first part will conform to an analytical approach comprising the matching of classification method support properties (i.e., what assessment criteria they intend to operate on) with the properties of the machines at the case company. This is undertaken to identify the most feasible classification method, and the intention is to enable this based upon transparent informative arguments motivating its feasibility. As previously mentioned, three classification methods have been selected to be subject to this exercise, the AHP, the VED, and the ABC analysis. See Table 5 to 10 for a summary of the properties previously outlined in Chapter 2.4. In Table 11, an illustration of the assessment criteria for each spare part unit for a given machine can be seen.

Table 11 - Classification method support properties.

The matching is then to be performed by assessing the criteria of the spare part management of selected machines according to Table 12 below. At the end of the columns a summation of the presence assessment criteria deemed crucial for the unit and can rate the importance of having a classification method. It also incorporates the most important assessment criteria, enabling for judging the feasibility of the methods. It is important to note that the assessment of the machines will be done by both interviews with machine operators and managers as well as through an analysis of the operational data in the production and maintenance system at the company.

(28)

21 Table 12 - Example of machine spare part unit criteria.

The second part of the application of method is to demonstrate the use of the classification method deemed as the most feasible one in an illustrative case at the case company. Based upon the demonstration, it is examined to what degree the classification method fulfills the requirement set below and provide decision information that may be acted upon.

3.4 Validation and reliability

The validation of this study is acceptable level because a mixed method approach is used during data collection, which is then compared with qualitative and quantitative data. The numerical methods and criterion used in this study need to be analyzed in a way that accurately reflects the views of the company and stakeholders. To make sure the criterion is chosen wisely, interview questions can be discussed with someone in the company that has sufficient knowledge about the subject. Finally, a group interview needs to be done with the participants who attended the previous interviews to receive verification. The validation can be called into question when author submits their subjective opinion about which criteria should be selected for each classification method.

The reliability of this study will be such that anyone can use it as an example in their company. Results will differ since stakeholders have different roles and purposes in the fir company, which means that preferences, criteria, and answers can change. The point is that anyone who utilizes this study in their field should not get the exact same result as this study but rather a result that will suit their working environment. All numeric data will be collected and calculated in Microsoft Excel to ensure there are no calculation faults when using the framework.

(29)

22

3.5 Ethical consideration

During this study, all participants and the case company will remain anonymous.

Only citations from the interviews will be published regarding confidential material. The purpose of this case study is to have anonymous participants which means any mass production factory with a similar problem to be solved can apply the results of this study to their organization. This study will remain neutral and will not provide information about the case company. In case any information about the machines or spare parts needs to be included they will be under different names. Only ideas and recommendations will be presented during the conclusion.

Therefore, any ethical issues during this case study will not be considered relevant to other cases.

3.6 Proof of concept

Proof of concept is a method to demonstrate a new application or technology. It is used in different fields such as business, engineering, and practice. Kendig (2015) described a proof of concept as a concept that works in a practical potential of a framework or prototype. The proof can be found through modeling, experimental investigation, and research. For this study, a proof of concept will be modeled by an experimental investigation to understand if the new methodology of framework is working or not.

(30)

23

4 Results

In this chapter the results of the study will be presented through collected data from interviews and results of calculated theoretical frameworks.

4.1 Interviews

Production area is a critical station where the customer’s product is being finalized and where bottlenecks may occur depending on how careful the maintenance and operators work. "The whole production line has a wide tolerance where the accuracy is not very high, but in this station, we are very careful about the quality". If the Producing machines stop for longer intervals they would lose downtime, lead time, and quality.

The production staff can manage normal production without one machine for a short period, but this would mean that they need to produce in smaller volumes and frequently to meet the demand volume. Machine without a backup and whose absence can halt the production line are considered to be critical. "Machines we only have one of and have a key role in production is a critical machine.". Most of the machines are linked which may result in a domino effect, if one machine stops working the following machines will inevitably suffer. The machines are pushed to their limits to manage with demand volume. Respondents concerns arises when they are explaining on how the machines may have a faster breakage when they are being pushed to their limits and how it may affect parts of the company. If production line due to machine stop functioning will require longer lead times to reach the costumers demand and may lose production. Another critical moment arises when the machine cannot be repaired within 10-20 hours or requested spare parts are not in stock in the warehouse.

The factory has machines that produce quality products and undergoes quality checks. Quality requirements are set by customers and the company regulation which needs to be achieved before the finished products can be shipped. Without achieved quality, the product cannot be sent to the customer. In case they are shipped away it would be resulting in reduced prices on the products and money loss for the company. Extra work might also be to analyze the rest of the products and understand reason for defects.

If two critical machines were to stop functioning due to a broken machine part, maintenance and property owners will jointly decide on which machine to start repairing. They will try to determine how important the machines are and whether the machine part affects another unit.

The respondents have different views on which machine had the most stops in 2019. One of the machines mentioned was Homogenization furnace 1, this machine has needed a lot of minor repairs. It is one of the critical machines for producing

(31)

24 quality products and has greater capacity for receiving volume of products. If this machine stops, delivery to the customer will also stop. This is due to the Producing machines, which means that if the Producing machines stop then the Homogenization furnace 1 will be affected. These two machines are prioritized meaning that their spare parts are critical, and quick delivery is required.

If one part of a machine stops working, it may be managed without affecting production, but it might also affect the rest (or other parts within the machine), meaning that some spares are dependent on others. Spare parts may age, if these parts are not caught in time, the machine may lose its functionality. Spare parts are considered critical if they affect the functionality of machine and if they belong to a critical machine. Another consideration is whether spares may be substituted and used to repair different machines. “Spare parts and critical spare parts may be separated from each other depending on where in the machine they are installed and what they are used for. You might say that they differ depending on their functionality”. Various factors determine whether a spare part is deemed a regular spare or a critical spare. “If it is possible to run with the broken part for a certain time, or if we can make our own solution then it’s deemed a regular spare. If we are completely dependent on it, with a long delivery time from the supplier, then it’s a critical spare."

When a machine stops a troubleshooter conducts an analysis to understand where the fault has occurred and what spare part needs replacing. If a spare is not in stock, there may be financial consequences and loss of production. Depending on which machine is affected, there may also be quality loss. The warehouse ensures spares are replenished daily, depending on the order point location of the spares.

However, the respondents explained that some spares are not registered, which means it takes longer determine whether the part is at the warehouse and enter information into the system. Some machines have long service life (durability) and may not have malfunctioned or caused a stop for several years. Thus, there has been no demand for those parts. It may also be that the wrong spare has been ordered and installed which leads to needless waste of work hours. When ordering a critical spare part, the aim is to have a fast delivery time, without price being a factor. Maintenance and warehouse only consider what downtime costs may arise if they continue operating the machine without its proper functionality. They believe these two issues may cost the company more than an expensive unit.

The company have no major problems with market availability or receive spare parts from suppliers on time. They have different suppliers for each spare but if the supplier does not have the spares, they either turn to other supplier or the supplier may check with their own contacts for help. As a result, they do not experience shortcomings in terms of suppliers. However, there is often a communication

(32)

25 problem between departments since they have different views on the matter which create longer delivery times to receive spare parts.

The case company does not have a systematic method for identifying critical machines and critical spare parts. Respondents declare they are not following up, reporting, or measuring any data. All critical machines and spare parts are managed by experience-based identification. In the maintenance system the machines are registered along with spare parts, but it is not clear which parts are critical and which are not.

(33)

26

4.1.1 Criteria observation

Table 13 is a summary of all the criteria that has been collected through interview data. Criteria were defined according the impact on spare parts and productivity stoppage. An explanation of each criteria has been done on how the participant validated the meaning of each criteria.

Table 13 – Criteria summary from interviews.

Market availability The number of suppliers the firm can buy spare parts from.

Lead time The volume of production that is

lost because of machine stoppage.

Durability The life span of a spare part.

Demand volume Volume the machine can produce

on customer demand.

Functionality One spare part can be suited to

different machines.

Quality Quality requirements on product.

Dependency One broken spare part can cause

another unit or machine to stop working.

Downtime The time elapsed between

machine stoppage and its restoration.

Unit volume The amount of spare parts is

needed in stock.

Availability The amount of spare parts in stock

until next machine stoppage.

Annual profit

The amount of profits the firm receives in a year. This can change in the event of a drop in quality or late delivery to costumer.

Delivery time The pf amount time takes for

spare parts to be delivered to the warehouse.

Costs The costs the firm might accrue in

case a longer work stoppage occurs.

Unit price The price of one spare part unit.

(34)

27

4.2 Experts design of mapping

The following table shows the results collected from interviews. The producing machine was selected by comparing the top 3 machine rankings. Further, analyses had to be carried out to understand which specific producing machine is the most critical one. The case company has three different producing machines, but two of them produce the same product. The participants explained that they identify the criticality of a machine by considering issues related to duplication and if there is risk of a stop in production if the machine fails. Thereby, producing machine 1 was chosen on the basis of the foregoing criteria. The property owner could only mention 8 critical units. Without these spare parts, there would be machine stoppage and a stop in production.

Table 14 comprises information that is within the purview of maintenance staff.

Demand volume, downtime, unit volume, annual profit, durability and costs are criteria with the highest total summation. This means those 5 criteria are the most relevant when understanding which classification method to use.

Table 14 – Framework result of Producing machine 1.

(35)

28

4.3 Technical design of mapping

Homogenization furnace 1 is the machine that had the most downtime during 2019.

The spare parts were chosen according to which units affected the machine’s efficiency and needed to be replaced. Table 15 shows the results that were taken from the case company downtime. Only 8 units were chosen from the considered collected units in Chapter 4.2. The summation shows unit price was the leading criteria. This means that all the 8 spare parts have a unit price listed in their purchasing system. Unit price is the information that the classification method can is used on.

Table 15 – Framework result of Homogenization furnace 1.

(36)

29

4.4 Properties and criteria matching

Table 16 gives results that are similar to Table 3 to 9 and from the previous research in Chapter 2. All criteria obtained in the interviews has been matched with the properties from the classification methods. This means that crosses have been added to the classification methods that referred to the identified criteria.

Table 16 – Summarized properties from previous works.

Table 17 shows what the mapping can look like by understanding the purpose of the classification methods and that of the criteria. All the crosses within the bracket refer to the mapping that is positioned according to the interpretation of the studies.

Meanwhile, only the cross is the result of previous research on Table 16.

Table 17 – Criteria concept according to the interpretation of the study.

(37)

30

4.5 Classification method support

The expertise design mapping gives results of the demand in volume, downtime, unit volume, durability, and costs in different classification methods. Since all the criteria are matched with different classification methods, the focus will be on the ones that match. Demand volume and costs are the two criteria that are mutual from the others. This means that AHP analysis and ABC analysis are the most profitable integrated classification methods, see Table 18.

Table 18 – Classification method mapping of producing machine 1.

The technical mapping in Table 19 shows that costs are the only relevant criteria.

The multicriteria classification method that can be suggested for use is AHP analysis and ABC analysis. By only working with historical data, an AHP analysis and ABC analysis can be used to identify the most critical spare parts. Working with these methods, the maintenance staff will have more control over what kind of costs might occur in case the unit stops working or spare parts might not be available in the warehouse.

Table 19 – Classification method mapping of homogenization furnace 1.

(38)

31

5 Analysis

Following chapter will present the analysis of the result and discussion of the experiment will be presented.

5.1 Criteria observation

From the observations made via the interviews, 13 different criteria were found regarding spare parts. These criteria are requirements that need to be met to be classified as critical but to also achieve the company's goals and increase annual profit. When analyzing the answers, criteria were discussed that could be used to determine the most critical machine for production. All criteria are connected to each other. The criteria are chosen in relation to the entire production line.

It is particularly important for the company to have market availability as they work through emergency situations. Having access to several suppliers with fast delivery is an advantage for warehouse staff and production planners. Other than delivery time, longer durability, and availability of spare parts are required to keep the machines running during a longer production period. The functionality of spare parts with parts from different machines is another requirement that count as critical criteria. To keep from falling behind in the maintenance of machines, the unit volume is regularly updated when it reaches the order point. The unit price only needs to be checked when the situation is not urgent. If the machinery malfunction, it is better to buy inexpensive spare parts; however, parts will probably not have long-term functionality and shorter durability.

Lead times are critical for the company, as these can result in a slowdown in production and a longer wait for customer to get their products. The respondents are aware that stops in production can contribute to less than economically viable results. Quality is also a requirement before products are shipped to customers.

Without good quality, there will be adverse economic results.

Time is valuable in production and downtime is limited as much as possible.

Machines are repaired quickly to continue regular production. Sometimes these machines are run with a spare part that is not working. A failed machine part working without any other problems is not critical. Yet this flawer part might be depended on other parts of the machine and need to be fixed sooner or later.

Maintenance staff have a common idea of what are considered to be critical parts and regular parts. They also agreed on the criteria and the definition of the criteria and they were similar to those in previous works. If the interview questions had been more in-depth, then more criteria could have been discovered.

During the interview, it was mentioned that there is a lack of communication between departments. Bad communication can cause delivery delays of spare parts and have negative effects on production. Whether a situation is deemed critical or

Integrated classification methods for spare parts