Assessment of critical success factors, barriers and initiatives of total productive maintenance (TPM) in selected Ethiopian manufacturing industries

As a result of the intense worldwide competition and ever-changing customer demands, many businesses have been driven to improve their performance through effective management programs, such as total productive maintenance (TPM) (Ahuja, 2009). Companies need to deploy high-performance manufacturing to stay competitive in the global economy, and maintenance management is a key function that supports effective production operations because increased maintenance function effectiveness leads to the development of key manufacturing competencies. The cost of operations and maintenance affects a company's ability to succeed (Chaurey et al., 2023). In general, the impact of insufficient and inefficient maintenance can determine the profitability and survival of a corporation. Furthermore, when this maintenance function is actively led, operational effectiveness improves (Al-refaie et al., 2022; Al-najjar, 2009). The importance of TPM in gaining and retaining a competitive edge has grown as worldwide competition has increased. To improve their competitive position, more and more companies are turning to proactive solutions like TPM. TPM plays a vital role in enhancing the global competitiveness (Toke and Kalpande, 2023).

TPM is a source of potential improvement for a company. It encompasses the whole organization, and when properly applied, it improves all areas of the business by increasing efficiency and improving overall performance. The relationship between people and machines is more productive when TPM is practiced (Kalpande and Toke, 2023). Being competitive in a market requires a high level of productivity and quality, and TPM is crucial to accomplishing these goals (Chaurey et al., 2023). TPM is becoming a critical strategic tool for businesses, and its significance is no longer limited to the manufacturing industry.

This study constructs a structural model to evaluate the interactions between the eight TPM pillars and overall equipment effectiveness (OEE) to assess the success of TPM pillar implementation, examine the linkages between them and study their implications on OEE. The findings give decision-makers a clear picture of the links between TPM pillars, highlight the most (and least) impactful pillars on OEE and point them in the right direction for improving Ethiopian manufacturing sector performance.

As far as the knowledge of the researchers, previous studies (Brown et al., 2002; Gupta et al., 2016; Sharma et al., 2012; Sharma et al., 2006; Singh et al., 2013; Tsang and Chan, 2000) mainly focused on TPM implementation without studying the existing status of TPM initiatives in manufacturing industries, little effort has been done to assess the overall initiatives of TPM implementation in manufacturing industries, identify the interrelationship between pillars of TPM and OEE, identify TPM pillars that have been best exercised and TPM pillars that needs more attention. Moreover, there are no recent studies done on Ethiopian manufacturing industries to identify barriers and critical success factors to TPM implementation. Therefore, this research is done to fill the gap stated.

Global competitiveness is one of the issues confronting the industrial sector today. OEE is one of the pillars of industrial competitiveness in the face of today's fierce global competition as it is one of the indicators of manufacturing performance. In the manufacturing industry, machine availability, performance and quality are crucial for increasing production and meeting or exceeding customer expectations. Maintenance activities must be well-managed to offer a high-quality product at a reasonable cost, allowing manufacturing companies to compete in the global market.

Ethiopian industrial enterprises have low production performance and high maintenance cost due to frequent machine malfunction (Lemma et al., 2013). As per the preliminary study and interview with employees of the organization in Ethiopian manufacturing industries:

1. Sudden breakdown of machines and production interruption is a common problem. The top managers follow the firefighting approach; the effort to prevent sudden breakdown is less.

2. There is no synergy between production managers and maintenance managers, the top management gives priority to production activities rather than maintenance related initiatives.

3. Most of the machines in the manufacturing industries are old where failure is a frequent problem.

4. As machines are too old the companies that manufactured forgot the machines and are not supplying spare parts to the market.

5. Experienced maintenance technicians are leaving the companies due to retirement age.

6. The effort to train and educate operators and maintenance technicians to operate and maintain the old and new machines is little.

7. There is a huge problem of spare parts related to the current currency starvation

All the aforementioned problems result in unnecessary production shutdowns, excessive maintenance costs, customer dissatisfaction due to delayed deliveries and poor product quality in the manufacturing industries.

Hence, following an appropriate maintenance policy/philosophy will alleviate these problems. As TPM is a maintenance philosophy targeting zero breakdowns, zero defect and zero accident through the involvement of all employees, it will help the manufacturing companies to solve the stated maintenance related problems. In addition to this; enlisting critical success factors and barriers of TPM implementation, identification of the weak and strong pillars of TPM will help companies to allot their limited resource to vital areas only.

1. Is there any initiative for TPM implementation in Ethiopian manufacturing industries?

2. What is the relationship between TPM pillars and OEE in the case companies?

3. What type of maintenance policy is commonly practiced in the case companies?

4. Which TPM pillars are best practiced in the case companies and which pillars of TPM are less addressed and need more attention?

5. What are the barriers and critical success factors toward TPM implementation in the case companies?

The general objective of this research is to investigate the critical success factors, barriers and initiatives of TPM implementation in selected manufacturing industries in Addis Ababa, Ethiopia.

In line with the general objective and research questions, the specific objectives of this research are:

1. To investigate the current initiatives of TPM implementation in Ethiopian manufacturing industries.

2. To investigate the relationship between TPM pillars and OEE in the case companies.

3. To identify the type of maintenance policy commonly followed in the case companies.

4. To identify TPM pillars best practiced in the case companies.

5. To identify TPM pillars less addressed so far and needs more attention in the case companies.

6. To identify barriers and critical success factors to TPM practice in the case companies.

In this section, definitions of Equipment maintenance, types of maintenance, the eight pillars of TPM and OEE have been discussed. The research hypotheses have been developed as well.

The practice of preserving a condition or situation or the state of being well-kept, is defined as maintenance. Almost every industry has a physical asset maintenance program. The goal of maintenance activities is to ensure equipment operational capability, detect problems and/or prevent functional failures (Joshua and Mathew, 2016). In industrial, commercial and residential settings, maintenance entails functioning checks, servicing, repairing or replacing necessary devices, equipment, machinery, building infrastructure and supporting utilities (Agency, 2016; Aigboje et al., 2019). This has evolved over time to include a variety of terms that indicate various cost-effective procedures for keeping equipment operating; these operations might occur before or after a breakdown (Peltier, 2011; Rastegari, 2015).

PvM is a method of equipment maintenance that entails replacing or repairing an asset regularly, regardless of its condition. Scheduled restoration and replacement are examples of PvM tasks. PvM is a sort of maintenance that is performed on equipment at regular intervals while it is still operational to prevent or minimize the chance of failure. PvM can be scheduled on a weekly, monthly or three-monthly basis (Trojan and Marçal, 2017).

Other types of maintenance that fall under the heading of PvM, in addition to the regular interval approach, include: time-based maintenance (TBM), failure-finding maintenance (FFM), risk-based maintenance (RBM), condition-based maintenance (CBM), reliability-centered maintenance (RCM), predictive maintenance (PdM) and prescriptive maintenance (PrsM).

CM is the process taken to restore an item to its original working state after failure or inadequacies were discovered during PvM or otherwise. After allowing an item to fail, a run-to-failure or CM plan restores its function. It is based on the notion that the failure is tolerable and that avoiding failure is either too expensive or impossible (Rastegari, 2015).

In addition to being the outcome of a planned run-to-failure technique, CM is the result of unforeseen breakdowns that were not prevented by PvM. When the implications of failure are low and repair is not required immediately, a run-to-failure technique can be used effectively.

In the chart of maintenance types, CM has been broken into two sub-types: deferred maintenance (DM) and emergency maintenance (EM) (See Figure 1).

TPM is a maintenance commitment that extends beyond preventive and predictive maintenance. TPM and total quality management (TQM) are comparable in certain aspects. The customer is the individual at the next level of the process, and all workers are involved in meeting customer demands. TPM interprets this as delivering the highest level of support and service to all equipment users. TPM considers a machine breakdown to be a fault, and it is dedicated to preventing breakdowns and malfunctions in the first place (Madani, 1996; Abhishek Jain, 2014; Prabowo et al., 2018).

As per the Japan Institute of Plant Maintenance (JIPM), TPM has eight pillars including autonomous maintenance (AM), focused improvement (FI), planned maintenance (PM), quality maintenance (QM), early equipment maintenance (EEM), education and training (EduT), office TPM (OTPM) and safety, health and environment (SHE) (see Figure 2).

The AM pillar is crucial. The term “total” in TPM refers to including everyone in the maintenance process (Chaurey et al., 2023). In a nutshell, the idea is that the machine's operators take on part of the maintenance responsibilities. This is a really smart and practical solution. Nobody knows the machine better than the individual who spends eight hours a day with it. To minimize failures and respond more rapidly when a specific defect is identified, AM allows operators to undertake minor maintenance operations on their own. This pillar covers the training of operators to handle initial maintenance needs as well as the responsibilities that are required on a regular basis to keep operations running smoothly (Trutkowski, 2016; Firdos Jahan and Quazi, 2014; Labiyi, 2019; Ngaich and Malviya, 2015).

FI is one of the framework's pillars, where TPM truly shines and brings significant value to the industry. It is centered on a team's collective effort to enhance the entire system. Their objective is to increase the equipment's capacity and efficiency while reducing the company's losses and waste. This pillar enhances efficiency by eliminating defects in products, processes and systems; it also improves safety performance by lowering risk factors in processes by assessing risk levels and taking appropriate steps; and it helps cross-functional teams identify and address recurrent issues (Labiyi, 2019; McKone et al., 2001; Prabowo et al., 2018).

EEM is a well-organized procedure that directs the machine's design and manufacture. It focuses on eliminating the challenges associated with machine operation and maintenance. Early equipment management attempts to provide new goods and processes with the shortest possible development time and the ability to mass-produce them. This pillar aids in the reduction of product and process setup time, the improvement of OEE and the delivery of large quantities of high-quality products (Ahuja and Khamba, 2008).

This pillar's goal is to increase equipment effectiveness and generate zero faulty items. Focusing on the preventive, breakdown and CM can help achieve this. Planned maintenance alters daily operations transforming maintenance operations from reactive to proactive (Chaurey et al., 2023). This action also improves dependability, maintainability and performance. With the use of recorded failure rates, we can design our maintenance program. The goals of this pillar are to have zero breakdowns, reduce mean time to repair (MTTR) and increase mean time between failures (MTBF). Planned maintenance aids in the reduction of breakdowns, the reduction of product defects, the improvement of machine efficiency, the improvement of machine dependability and maintainability, the reduction of unplanned downtime and the growth of production uptime (Adesta et al., 2018; Firdos Jahan and Quazi, 2014; Ngaich and Malviya, 2015).

The basic idea behind QM is to provide a framework for detecting and preventing errors in manufacturing processes. Zero breakdowns lead to defect-free output, often known as zero quality faults. Observing the changes in functioning machines can assist us in preventing machine failure defects. Everyone is in charge of maintaining ideal circumstances and achieving zero faults. By enhancing the product's quality, QM lowers the cost of quality, minimizes poor quality wastes and reworks and lowers customer complaints (Adesta et al., 2018; Firdos Jahan and Quazi, 2014; Ngaich and Malviya, 2015).

The application of TPM techniques to administrative operations is the focus of this pillar. This pillar's goals include achieving zero function losses, establishing efficient offices and implementing service support functions for industrial processes. This pillar has several advantages, including better utilization of workspace, reduction of repetitive tasks and increased inventory levels across the whole supply chain. Administrative and overhead costs have been reduced. Due to logistics, there has been a decrease in the number of office machine breakdowns and customer complaints, as well as a decrease in the cost of emergency dispatches and purchases (Adesta et al., 2018; Prabowo et al., 2018).

Maintaining a safe and healthy working environment is the focus of this pillar. This pillar's goal is to have no accidents, no health problems and no fires. As a result, workers must be able to carry out their duties in a safe workplace free of health hazards. This pillar's implementation avoids possible health and safety issues. This pillar guarantees that all employees operate in a safe and accident-free workplace. It enhances the motivation of employees. Employee attitudes about work alter substantially in a safe atmosphere, resulting in increased productivity, quality and delivery performance (Agung and Siahaan, 2020; McKone et al., 2001).

The most crucial aspect of accomplishing the TPM goal is training. All operators, supervisors and managers are affected. Operators learn how to maintain the machine and spot any faults. The operator learns proactive and preventive machine maintenance practices. Training and education guarantee that employees are properly trained to do their jobs; managers identify the skill, which aids in the effective implementation of TPM by the organization's goals and objectives. The appropriate kind of practice and training raises skill levels and constant repetition produces flawless performance (Chaurey et al., 2023).

All personnel's abilities and performance improve as a result of education and training. Human potential wastes a lot of money in a company if it is not developed. We may have multi-skilled employees through training and education, allowing the operator and supervisor to perform more efficiently, effectively and autonomously (Firdos Jahan and Quazi, 2014; Kelly et al., 2002; Prabowo et al., 2018).

Manufacturing productivity is measured by OEE. It expresses how much of the equipment's full capability is being utilized (in %). The measure is largely determined by assessing three OEE factors: availability, performance and quality. OEE is not just an element of the TPM concept, but it may also be utilized independently as a machine or process performance evaluation. The goal of OEE is to continuously enhance and maintain the equipment in terms of effectiveness and efficiency (Dreher et al., 2019; Al-refaie et al., 2022; Ngaich and Malviya, 2015). OEE is a component of the industrial efficiency improvement process (Vairagkar and Shyam, 2015; Bamber et al., 2003; Yimam Ali and Ali, 2019).

In this study, a conceptual research model shown in Figure 3 was developed. The model hypothesizes that the implementation of TPM pillars leads to OEE improvement and if TPM pillars are implemented, it shows the presence of good TPM implementation initiative in the organization. Moreover, this model based on the information from previous literature hypothesized that EduT have direct and indirect effect on OEE.

In studying the relationship between EduT pillar, other TPM pillars and OEE, the dependent and independent variables should be identified. In the conceptual model, EduT pillar is defined as an independent variable, while OEE is classified as a dependent variable. The AM, FI, EEM, Planned Maintenance (PM), QM, OTPM, and SHE constructs are endogenous and have dual relationships as both independent and dependent. They are dependent constructs because they are predicted by EduT, but they are also independent constructs because they predict OEE. The path relations of the variables are shown in the conceptual model (See Figure 3).

This section discusses the conceptual research model, sample size, sampling technique, research instrument, study area, time frame, data analysis method, data collection and processing procedures.

The area of this study is Addis Ababa, Ethiopia. This area is selected as it is the one with many manufacturing industries in the country.

The sampling technique applied for this research is purposive sampling. The key informants for this study were selected based on maintaining the appropriate mix of individuals related to the research title from 18 manufacturing industries. The selection of the 18 manufacturing industries was based on factors including the machine-intensive operations that make the implementation of TPM more relevant in these industries, the company's willingness to share information with the researchers about TPM, the company's representativeness in its sector and the ease of accessing the firm's operations and data. Metal industries, food processing industries, printing industries, bottling and packaging industries, chemical industries and tire manufacturing industries were all included in the list of selected industries. To determine the sample size required for analysis using PLS-SEM, the minimum R-squared table produced by (Kock and Hadaya, 2018; Priyanath et al., 2020), has been employed. Three elements are required to determine the minimum sample size using this method. The first element is the maximum number of arrows pointing at latent variable. The used significance level is the second element and the third element is the minimum R² in the model. In this research model, there are seven arrows pointing to a construct (see Figure 3). Using a 5% significance level and assuming the minimum R² value to be 0.25 and referring Table 1 this research required 80 samples to reach 80% prediction power for identifying R² values of at least 0.25 with a 5% chance of error. The researchers feel that the 102 data that have been gathered go well beyond the necessary minimum.

The primary research instrument that was used to collect relevant data concerning the practice of TPM and OEE from manufacturing industries in Addis Ababa, Ethiopia, was a close ended questionnaire and semi-structured interview questions. The questionnaire was adapted from previously used well tested questionnaires (Al-refaie et al., 2022; Ahuja, 2009) in the literature concerning TPM pillars and OEE. The researchers conducted semi structured interviews with industry mangers, operators and maintenance technicians to triangulate the finding of the questionnaires.

The questionnaire was filled by technical managers, general managers, maintenance technicians and operators of the company, who have experiences of more than two years and above. During the data collection, the purpose of the study was made clear and the consent of the respondents was assured with official letters written from the School of Mechanical and Industrial Engineering, Addis Ababa Institute of Technology, Addis Ababa University, Ethiopia. Before the data collection, the questionnaire was peer reviewed and pilot tested in five industries. Based on the feedback obtained, the questionnaire was modified and improved, such as eliminating redundant questions or improving the clarity of the questions. At last, the reliability of the measurement model was tested using SPSS version 23 software the result shows that Cronbach's alpha value greater than 0.70 and above, which is above the acceptable cut-off point (Cronbach, 1951). In the end, the final questionnaire (see  Appendix 1, Table A1) was distributed to the target sample of 130 respondents. Among distributed questionnaires 102 collected showing the response rate to be 78.46%.

The study was conducted in Addis Ababa, Ethiopia. It has considered the selected 18 manufacturing industries in Addis Ababa, Ethiopia. The data collection was conducted in a time frame of February 2022–April 2022.

After the construction of the research's conceptual model (Figure 3), the questionnaire for data collection was then created. The questionnaire consists of three sections. Section 1 involves general data, such as personal information, type of maintenance policy and company size information. Section 2 deals with questions related to TPM pillars and OEE. Section 3 contains questions related to opinions of respondents, regarding implementation of TPM in their company and critical barriers and success factors of TPM implementation. The questionnaires in section 2 used a five-point Likert scale from Strongly Disagree to Strongly Agree. A self-administered questionnaire was distributed to 18 manufacturing companies in Addis Ababa, Ethiopia. A total of 102 valid responses were recorded. The data were then processed using partial least squares structural equation modeling (PLS-SEM) method. PLS-SEM is one of the most popular multivariate data analysis techniques and has a remarkable capacity for working with small samples (Hair et al., 2021), which is the justification for choosing it for data analysis. Moreover, PLS-SEM is the preferred method even when the study object does not have a well-developed theoretical base, particularly when there is little prior knowledge of causal relationships. Unlike covariance-based structural equation modeling (CB-SEM) and multilinear regression analysis, PLS-SEM does not require a large sample size, a specific assumption about the distribution of the data, or even the missing data (Hair et al., 2014). Users with small sample sizes and less theoretical support for their research can apply PLS-SEM to test the causal relationship. The algorithm of PLS-SEM is different from the common structural equation modeling (SEM), which is based on maximum likelihood. When the sample size and data distribution of research can be hardly used by a common SEM, PLS-SEM has a more functional advantage (Fan et al., 2016) (Hair et al., 2014).

Descriptive statistics were initially used to analyze the data. Additionally, given that there were numerous independent factors and a dependent variable in this study, path analysis using Partial Least Square Structural Equation Modeling (PLS-SEM) was used to test the main effect and mediation hypotheses. To this effect, the data were analyzed using SPSS Version 23 and SmartPLS 3.0 software.

The descriptive data which includes gender, age, educational level, work experience, position of the respondents, company type and the company size are shown in Table 2.

As shown in Table 2, the majority of respondents (96.08%) were males, while only (3.92%) were females, indicating that the proportion of females in this study is too low. The bulk of respondents (65.69%) had first-level degrees, while (34.31%) had master's degrees. The majority of respondents (32.35%) have worked in the case companies for more than 15 years, followed by 26.47% who have worked there for 6–10 years, 26.47% who have worked there for 11–15 years and 14.71% who have worked there for 2–5 years. It was done on purpose to include technical managers, maintenance technicians, general managers, production managers and operators among the positions of the respondents as these respondents may have more reliable information about TPM. The sizes of all the selected manufacturing organizations were large.

As shown in Figure 4, a large number of respondents (85) indicated that breakdown maintenance is the most commonly employed maintenance strategy in their company, followed by TBM (56), CBM (28), PdM (10), TPM (7) and PrsM (1). According to the responses, the majority of the case companies use breakdown maintenance as their primary maintenance strategy.

According to Figure 5, the majority of respondents (61.8%) believed that TPM was not adopted in their company, 27% believed that it was only partially implemented, and only 11.2% said that TPM had actually been implemented in their organization. Based on the responses, TPM is not implemented in most of the case companies.

As shown in Table 3, the main implementation challenges for TPM are broken down into the following categories: behavioral, strategic, technical, human and cultural, operational and financial, supporting the findings of (Chaurey et al., 2023; Ahuja and Khamba, 2008; Panneerselvam, 2012; Tyagi et al., 2013). In addition to this, critical barriers are ranked under these categories based on the opinions of the respondents.

Table 4 shows the ranked critical success factors of TPM implementation as the per the response rate of the key informants. The result supports the finding of (Ahuja and Khamba, 2008; Panneerselvam, 2012; Tyagi et al., 2013).

As indicated in Table 5 the mean value for the constructs (TPM pillars) is considered low, since the corresponding mean value is lower than the mid-point of the response scale (i.e., three for a five-point Likert scale). The low mean values for all TPM pillars indicate that there is low initiative toward implementation of TPM in the case companies.

The association between the TPM pillars and OEE is significant at the 0.01 level for each correlation value (see Table 6). This suggests that OEE will increase as TPM pillar deployment increases and vice versa. Additionally, there is a strong and positive association between the TPM pillars.

The first step in PLS-SEM analysis is to evaluate the measurement model. The purpose is to determine how well the items load on the hypothetical-defined construct. The assessment of outer model involves: the examining of reliabilities of the individual items, reliability of each latent variables, internal consistency, convergent validity and discriminant validity (Hair et al., 2017).

Composite reliability (CR), individual indicator reliability and average variance extracted (AVE) are all used to evaluate the measurement model's internal consistency and convergent validity. Additionally, the Fornell–Larcker criterion and cross-loadings are used to assess discriminant validity.

According to a common rule of thumb for indicator reliability, a latent variable should explain a significant part, usually at least 50%, of each indicator's variance (Hair et al., 2013). Therefore, the outer loading of an indicator should be more than 0.708, because that value squared (0.708)² equals 0.50. Except for EduT3 (0.699), all the indicators for the constructs in this work were well above the minimum acceptable level for outer loadings (see Figure 6 and Table 7).

The most common measurement used for internal consistency is Cronbach's alpha and CR, in which it measures the reliability based on the interrelationship of the observed items. In PLS-SEM, the values are organized according to their indicator's individual reliability (Hair et al., 2017). The values range from 0 to 1, where a higher value indicates higher reliability level. Cronbach's alpha and composite reliability value > 0.70 is acceptable (Cronbach, 1951; Hair et al., 2017). As shown in Table 7, Cronbach's alpha values for all constructs are > 0.70 and the composite reliability (CR) of all variables are > 0.70, showing the internal consistency of the measurement items.

Convergent validity is the assessment to measure the level of correlation of multiple indicators of the same construct that are in agreement. To establish convergent validity, the factor loading of the indicator, CR and the AVE have to be considered (Hair et al., 2017). The value ranges from 0 to 1. AVE value should exceed 0.50, CR and the indicator's outer loadings should be higher than 0.708 so that it is adequate for convergent validity (Henseler et al., 2009; Ab Hamid et al., 2017; Hair et al., 2013). As shown in Table 7, all values of AVE are greater than 0.5. The factor loading and CR values are >0.708 showing the convergent validity of the measurement model.

Discriminant validity is the extent to which a construct is truly distinct from other constructs by empirical standards. Thus, establishing discriminant validity implies that a construct is unique and captures phenomena not represented by other constructs in the model. The Fornell–Larcker criterion is a conservative approach to assessing discriminant validity. It compares the square root of the AVE values with the latent variable correlations. Specifically, the square root of each construct's AVE should be greater than its highest correlation with any other construct. This criterion can also be stated as the AVE should exceed the squared correlation with any other construct. The logic of this method is based on the idea that a construct shares more variance with its associated indicators than with any other construct (Hair et al., 2013) (Hair et al., 2017). As shown in Table 8: the diagonal (italic) values are $AVE$ and the other values are correlations. The square root of each construct's AVE is greater than its highest correlation with any other construct. Therefore, discriminate validity criteria are fulfilled.

The structural model is evaluated by examining at its predictive capabilities as well as the relationships between the constructs. The significance of the path coefficients, level of R² values, f² effect size, predictive relevance Q² and q² effect size, are the key criteria for evaluating the structural model in PLS-SEM. According to (Hair et al., 2013) assessment of structural model has five steps including: assessment of structural model for collinearity issues, assessment of the significance and relevance of the structural model relationships using structural model path coefficients, assessment of the level of R², assessment of the effect sizes f², assessment of the predictive relevance Q² and the q² effect sizes.

Before conducting the analyses, the structural model must be examined for collinearity. The path coefficients might be biased if the estimation involves significant levels of collinearity among the predictor constructs. If the level of collinearity is extremely high as indicated by a variance inflation factor (VIF) value of five or higher, one should consider removing one of the corresponding indicators (Hair et al., 2013). As shown in Table 7, all constructs have a VIF value of less than five showing there is no collinearity issue. OEE2 has a relatively highest value of VIF (4.480) but still within the limit.

After running the PLS-SEM algorithm, estimates are obtained for the structural model relationships, which represent the hypothesized relationships among the constructs. The bootstrapping result shows that except the two paths (EEM→OEE and OTPM→OEE), all paths have a positive relationship with their dependent variable; however, not all variables are statistically significant (See Figure 7 and Table 9). As indicated by the bootstrapping results of PLS-SEM in Table 9: the direct effect of the five pillars of TPM namely AM, EEM, PM, OTPM and SHE was not significant. The total indirect effect of EduT pillar on OEE was found to be significant. Moreover, the direct effect of two pillars of TPM namely FI and QM were found to have a significant direct effect on OEE. The hypotheses H2, H5, H8b, H8c, H8d, H8e, H8f, H8g, H8h and H9 were accepted and H1, H3, H4, H6, H7 and H8a were rejected (See Table 9).

As shown in Table 9, the direct effect of EduT on OEE is not significant. Accordingly, hypotheses H8a was rejected. But the total indirect effect of EduT on OEE as mediated by AM, EEM, FI, OTPM, PM, QM and SHE is positive and significant supporting hypothesis H9 (see Table 10).

The coefficient of determination (R² value) is the metric that is most frequently used to assess the structural model (Hair et al., 2013; Hair et al., 2021; Henseler et al., 2009). This coefficient, which is determined as the squared correlation between the actual and anticipated values for a certain endogenous construct, serves as a gauge of the model's predictive efficacy. The coefficient shows how the endogenous latent variable is affected by the combined impact of the exogenous latent variables. The amount of variance in the endogenous constructs that is explained by all of the exogenous constructs linked to it is also represented by the coefficient, which is the squared correlation of the actual and predicted values.

The R² value ranges from 0 to 1 with higher levels indicating higher levels of predictive accuracy. In general, R² values of 0.75, 0.50 or 0.25 for the endogenous constructs can be described as substantial, moderate and weak, respectively (Hair et al., 2013; Hair et al., 2021; Henseler et al., 2009). Table 11 shows the R² values for all endogenous variables. The TPM pillars collectively explain 60.2% of the variance of OEE, while the remaining 39.8% is explained by some other variable, according to the OEE's significant R² value (0.602). Similar to AM, EEM, FI, QM and SHE all have R² values that are near to moderate (0.320, 0.437, 0.318 and 0.326, respectively). The R² values for OTPM (0.299) and PM (0.288) are weak. These results demonstrate that the EduT pillar of TPM may predict differences in 41.9% of AM, 32.20% of EEM, 43.7% of FI, 31.8% of QM, 32.6% of SHE, 29.9% of OTPM and 28.8% of PM. Other factors are responsible for the remaining percentages for each TPM pillar.

The change in the R² value when a certain exogenous construct is excluded from the model can be used to assess if the excluded construct has a significant impact on the endogenous constructs in addition to analyzing the R² values of all endogenous constructs. The f² effect size is the name given to this metric. According to the guidelines for calculating f², values of 0.02, 0.15 and 0.35, respectively, correspond to the minor, medium and large effects of the exogenous latent variable (Salkind, 2012).

The f² values of AM→OEE (0.002), EEM→OEE (0.008), OTPM→OEE (0.016), PM→OEE (0.003) are less than 0.02 showing the small effect of these pillars if we remove them from the model. The f² values of QM → OEE (0.086) and SHE → OEE (0.024) shows the variables have medium effect on OEE. The f² value of EduT → AM (0.719), EduT → EEM (0.474), EduT → FI (0.772), EduT → OTPM (0.425), EduT → PM (0.402), EduT → QM (0.468) and EduT → SHE (0.484), shows that EduT will have a large effect on the endogenous corresponding variables if removed from the model (See Table 12).

Researchers should look at Stone–Geisser's Q² value in addition to the size of the R² values when determining the predictive accuracy (Geisser, 1974). The predictive relevance of the model is indicated by this metric. The external constructs have predictive relevance for the endogenous construct under examination, as indicated by Q² values larger than 0. As a relative measure of predictive importance, the values of 0.02, 0.15 and 0.35, respectively, imply that an external construct has a minor, medium or strong predictive relevance for a certain endogenous construct (Q²) (Geisser, 1974) (Hair et al., 2013). Table 13 shows that all variables have medium predictive relevance.

The key constructs that are most relevant to explaining the endogenous latent variable(s) in the structural model can be determined by interpreting these findings. As a result, Table 13 shows that AM, EEM, FI, OTPM, PM, QM and SHE have medium relevance to the endogenous variable OEE.

The main outcome of this study offers empirical support for the case companies' TPM implementation initiatives. From the findings of descriptive statistics and multivariate analysis using PLS-SEM, the TPM implementation initiative of the case manufacturing industries is at its earliest stage. To reach the OEE level for the world-class production, a greater effort needs to be done.

The research also discloses the barriers and critical success factors of TPM implementation in the selected manufacturing companies. Regarding the implementation of the eight TPM pillars and their effect on OEE, only three pillars namely FI, QM and EduT have significant effect on OEE of the selected companies. The other five TPM pillars—AM, EEM, Planned Maintenance (PM), OTPM and SHE—had no appreciable impact on the situation. Although EduT have a small direct impact on OEE, they have a considerable indirect impact on OEE through the mediation of AM, FI, PM, EEM, QM, SHE and OTPM. The fact that OEE's R² = 0.606 indicates that the total effect of TPM pillars accounts for 60.6% of OEE's explanation, with other variables accounting for the remaining 39.4%.

The originality of this study is in its thorough analysis of the connection between TPM pillars and OEE in manufacturing industries, particularly in Ethiopia, by creating a unique conceptual model that is extremely distinct from previous studies. This study filled up the gaps left by earlier research by examining important data about critical success factors and barriers of TPM implementation as well as by identifying the weak and strong pillars of TPM in the instance manufacturing companies. Managers of the organizations can use this knowledge to focus their limited resources on the particular issues that need their attention. Future research in the topic can be built on the findings of this study as well.

TPM is not a quick fix. It necessitates a change in both the company's and employee's attitude and their values. Bringing about this change requires time, long-term planning, the dedication of top managers, as well as EduT. The management of the organizations should focus on:

1. Eliminating the barriers of TPM implementation.

2. Taping to critical success factors facilitation of the implementation of the weak pillars.

3. Maintaining and improving TPM pillars that have good implementation.

4. Identifying the skill gaps and filling the gap through appropriate EduT programs.

Despite the fact that great effort has been done to maintain the quality of this study and the study investigated helpful information to the practitioners and researchers, this study is not without limitations. First, the limitation of this research is the small sample size. Due to limitation of time and budget, this research focused on 18 purposely selected manufacturing industries in Addis Ababa, Ethiopia. Accordingly, it is difficult to generalize the research finding to the other sectors. Second, the manufacturing performance KPI used in this study is only OEE. Future studies should focus on enlarging the coverage of the case companies to get generalizable result. Other measures like MTBF, MTTR and mean down time (MDT) can be included to the manufacturing performance measurement KPI (OEE).

Addis Ababa University.

Institute of Technology.

School of Mechanical and Industrial Engineering.

Questionnaire.

Dear respondent,

Firstly, we express our gratitude for your kind support and sharing your knowledge while answering a few questions below.

Nowadays, extensive study is being conducted to identify various strategies and processes that might help a company enhance its quality and productivity to cope up with the global competition. One such method is Total Productive Maintenance (TPM). It is a Japanese philosophy that focuses on obtaining ZERO breakdowns, ZERO defects and ZERO accidents by properly maintaining equipment throughout its lifetime. Total productive maintenance (TPM) is a strategy that operates according to the idea that everyone in a facility should participate in maintenance, rather than just the maintenance team only. TPM integrates all areas of an organization.

Dear respondent, we are conducting research entitled “Assessment of critical success factors, barriers, and initiatives of Total Productive Maintenance (TPM) in Selected Ethiopian Manufacturing Industries”. Your responses in this regard: shall help us to complete this research in efficient way, will be strictly kept confidential and used for academic purpose only. The results of this study will be reported only in aggregate form (no individual names and company will be reported). You can also be informed of the outcome of the research study, if you desire.

It will be a great favor for us and the case companies if you can fill and submit this survey as early as possible.

If you have any doubt, please don't hesitate to contact us at:

Mulatu Tilahun

E-mail: MTBT2017@gmail.com

Mobile: +2519282879 66.

Many thanks in anticipation for all your help, time and effort without which this research is not effectively possible!

With best regards!