KPI tree - a hierarchical relationship structure of key performance indicators for value streams

© 2023 Author(s). This is an open access article licensed under the Creative Commons Attribution (CC BY) License (https://creativecommons.org/licenses/by/ 4.0/). 175 ARCHIWUM INŻYNIERII PRODUKCJI PRODUCTION ENGINEERING ARCHIVES 2023, 29(2), 175-185 PRODUCTION ENGINEERING ARCHIVES ISSN 2353-5156 (print) ISSN 2353-7779 (online) Exist since 4th quarter 2013 Available online at https://pea-journal.eu KPI tree - a hierarchical relationship structure of key performance indicators for value streams Alberto Bumba1, Manuel Gomes1, Cristiano Jesus2,3 , Rui M. Lima3 1 Bosch Car Multimédia Portugal /Assembly INF, PRO, CU, CC (BrgP/MFE21) (BrgP/MFE21), 4701-970, Braga, Portugal, 2 CiTin – Industrial Technology Interface Centre, Advanced Production Systems Department, 4970-786, Arcos de Valdevez, Portugal, 3 ALGORITMI Research Centre / LASI, Department of Production and Systems, School of Engineering, University of Minho, 4800-058 Guimarães, Portugal, *Correspondence: [email protected] Article history Received 18.06.2022 Accepted 09.12.2022 Available online 08.05.2023 Abstract Performance Measurement Systems (PMS) have been a potential answer to problems related to production systems monitoring, allowing the management and manipulation of data collected at various levels in organizations. PMS can be defined as a group of indicators in an information system. There are several types of PMS, however, the relationship between indicators in a PMS is still an issue that needs to be explored, as the KPIs in a production system are not independent and may have an intrinsic relationship. The purpose of this paper is to present a multilevel structure and its intrinsic structural relation for managing and analysing KPIs for a value stream production system. This hierarchical structure has different KPI levels such as Improvement KPIs, Monitoring KPIs, and Results KPIs or KPR (Key Performance Results), intrinsically related from the strategic levels to the operational levels. This provides a useful tool for the management of production systems, being used to analyse, and support the organization's continuous improvement processes. Keywords Performance Measurement System KPI – Key Performance Indicators Value stream performance Lean Production Continuous Improvement DOI: 10.30657/pea.2023.29.21 1. Introduction With the increase in competitiveness and market demand, companies are obliged to adopt practices that aim at cost optimization, increase in quality, and renewal of products and their useful life, to survive the conjuncture (Suzaki, 2017). With the emergence of the Lean Production philosophy, based on the Toyota Production System (TPS), whose goal is to increase value for the client while simultaneously striving for the elimination of waste, companies have gained more concepts and tools to face new challenges (Womack, Jones, and Roos 1990). However, the survival of companies has been related to long-term competitiveness, i.e., companies must guarantee a production system characterized by high performance in terms of reliability, sustainability, flexibility, and productivity (Ante et al. 2018). According to Mejjaouli and Babiceanu (2014), manufacturing companies characterized by long-term competitiveness issues are subject to more complex problems, which in most situations have to do with compliance, low stocks, uncertainties in demand, standardization of processes, and product development complexity. With the rise of industry 4.0, it became easy to collect data on machines, emphasizing quality, and avoiding flaws in the production process. However, this paradigm creates the opportunity for great flexibility and competitiveness in production systems, but at the same time requires a high level of system control, which depends on the ability to measure, monitor, and evaluate the system parameters (Lu, 2017). According to Braz, Scavarda, and Martins (2011), one of the pillars that makes it possible to face the challenge of monitoring the performance of production systems, is the implementation of a robust system to control and monitor the entire production system. Aikhuele, Ansah, and Sorooshian (2017) consider that performance measurement is an integral part of a planning and control system that cannot be treated in isolation, but rather as part of a strategy to evaluate actions taking into account ALBERTO BUMBA ET AL. / PRODUCTION ENGINEERING ARCHIVES 2023, 29(2), 175-185 176 ARCHIWUM INŻYNIERII PRODUKCJI efficiency and effectiveness. Therefore, a production manager can evaluate performance through the analysis of KPIs, which allows quantifying the efficiency and effectiveness of actions both in part and also in the entire production process without losing sight of general aspects such as strategic directives, customer satisfaction, and other intangible parameters (Ante et al., 2018a). Therefore, manufacturing industries have incorporated several systems to evaluate the performance of the production processes, called Performance Measurement Systems (PMS). A PMS consists of a set of metrics capable of quantifying the efficiency and effectiveness of the production processes (Neely, Gregory, and Platts 1995). In a PMS, the strategic objective is defined according to the needs of the company, then, each objective is supported by a detailed set of indicators that contribute to achieving the strategic objectives. These indicators are called Key Performance Indicators (KPIs), which consist of a quantifiable set of measures in a PMS, which reflect the critical success factors of the company or a particular activity (Kang et al., 2016). KPIs plays a crucial role in the study and improvement of the production system performance, according to International Standard ISO 22400-1 (2014) and International Standard ISO 22400-2 (2014) report, it presents a set of 34 KPIs along with their contexts and content. Some of these KPIs are not independent, with an intrinsic relationship between them. Therefore, for the effective use of KPIs for production control or continuous improvement, understanding the relationships between them is very important. In a production system, once a set of KPIs is defined in a PMS, each parameter reflects a facet of the system's performance. Since different variables of performance are not independent and cannot be separated, KPIs also have mutual relationships. Some KPIs can be correlated positively or negatively. Some can be obtained and replaced by others. To effectively use KPIs for continuous improvement (CI) or production control, it is important to understand these relationships (Kang et al., 2016b). Kang et al. (2016) stated that the investigation of KPI relationships relies mainly on statistical data-based approaches. This method identifies positive or negative correlations between KPIs. However, it may fail to find intrinsic connections and managerial insights. In addition, data collected from different companies can lead to substantially different results. Therefore, a new approach to discovering KPI relationships through intrinsic implications needs to be developed. To achieve this, KPIs need to be properly arranged at different levels, which means, a hierarchical structure must be developed. There are several hierarchical structures developed, Cross and Lynch (1988) proposed the SMART (strategic management and reporting technique), a four-level performance pyramid that connects company strategy with operations through the hierarchy, transforming objectives from up to measures from bottom. Fitzgerald et al. (1991) developed the Result and Determinant Framework. The structure divides measures into two categories: results (measures that are actions result, for example, competitiveness, and financial performance) and determinants (measures that measure actions that lead to certain results, for example, quality, flexibility, use of resources and innovation). Kaplan and Norton (2005) developed the Balanced Scorecard, the well-known performance measurement system. It is a balanced performance measurement system; it contains both financial and non-financial measures. It sees the business comprehensively from four different point of views (customer, financial, innovation and learning, and internal processes). Ruano Pérez et al. (2018) and Perera and Perera (2019) identify the growing importance of hierarchical models that have been drawing the attention of practitioners and also researchers, given the growth of work done in this area. Despite these efforts stated above, there is still space for improving the knowledge of the intrinsic relationships of KPIs in production systems (Ante et al., 2018a). Thus, this article aims to improve the knowledge of the intrinsic relationships of KPIs in production systems, presenting details of a value stream performance measurement system, based on a KPI tree, a hierarchical structure used to describe and relate KPIs from the strategic to the operational level. The KPI tree for this company was presented by Ante et al. (2018) and the current submission increases the knowledge related to the specific utilization of such a tool for measuring the performance of a value stream. Thus, this tool is structured considering a lean production system, organized by value stream, to collect data at a very specific level. The system in this study belongs to the Bosch group, located in BragaPortugal. The KPI tree is composed of several levels of indicators, and based on this, the relationship between levels and between indicators within the levels is explored. In addition, the usefulness of using the KPI tree in continuous improvement projects will be presented. 2. Literature review Performance measurement systems are mainly based on financial measures, are strongly results-oriented and have a focus on past actions since they can describe actions or decisions only after they are applied (Hatzigeorgiou and Manoliadis, 2017; Staedele et al., 2019; Susilawati, 2021). This idea is reinforced by Peñaloza, Formoso, and Saurin (2017), in an analysis of PMS in the area of security, refer that the evaluation indicators use a retrospective or evaluation perspective of past conditions based on statistical data. Roth, Deuse, and Biedermann (2020) call static PMS those that collect and evaluate indicator data, but their statistical instruments are not sufficient to consider the dynamic behaviour of the organization, and therefore cannot evaluate trends, and reflections of variability, waste, etc. Therefore, these authors advocate a framework that advocates a dynamic system of performance measurement structured from the vertical and horizontal integration (inter and intra linkage) of different dimensions composed of various quantitative and qualitative indicators. Based on an extensive literature review investigation, Aikhuele, Ansah, and Sorooshian (2017) identified several PMS and concepts and, as a synthesis, the authors refer that ALBERTO BUMBA ET AL. / PRODUCTION ENGINEERING ARCHIVES 2023, 29(2), 175-185 ARCHIWUM INŻYNIERII PRODUKCJI 177 most of these systems are limited to a certain delimited set of values focused on a few aspects of lean manufacturing principles. For example, a proposed PMS for British manufacturing companies, they are based on 5 lean enablers, namely supplier relationship, lean management, lean workforce, process excellence and customer relationship, and considers supplier delivery, management culture and process optimization as performance indicators. Several authors hold the view that more effort needs to be put into developing a model that evaluates all lean principles (Carneiro et al., 2017; Khaba and Bhar, 2017), going beyond traditional indicators such as meeting delivery deadlines, financial indicators, and other specifications, and include factors such as value maximization, waste minimization, cycle time reduction, and production flow stability improvement, which causes the PMS to accommodate indicators that are characterized by a gradation between the quantitative and the qualitative. Chiarini and Vagnoni (2015) stated that the TPS-lean PMS is usually based on funded limited set of measures that preys on the speed and frequency of the performance indicator measurement process. However, in a study conducted by the authors on the world-class manufacturing concept developed by Fiat, they identified that the company has developed its own PMS that integrates the strategic objectives, safety, quality, environment, and energy management aligned with other goals established in the strategic planning. Although the performance indicators are integrated into a single system, they are designed in such a way that safety and quality cannot have trade-offs with costs or other strategies. Therefore, it is an articulated model of performance indicators. The increasing complexity of goods and services production systems, therefore, has provoked researchers and practitioners to seek an articulated solution for performance evaluation. Several authors (Hatzigeorgiou and Manoliadis, 2017; Staedele et al., 2019; Susilawati, 2021) consider that with the emergence of the Lean Production philosophy, performance evaluation has gained new challenges and greater complexity as objectives such as waste reduction, variability reduction, and simplification of operations have become pursued, and new forms of evaluation such as benchmarking have been adopted to address this need for more comprehensive evaluation. A line of thought related to PMS approaches is rooted in the quality management area, being two examples of those the European Foundation for Quality Management (EFQM) Excellence Model and Balanced Scorecard (BSC) (Hatzigeorgiou and Manoliadis, 2017). The BSC proposes a balanced set of measures that provides top management with an overall understanding of the business (Olivella and Gregorio, 2015); however, the authors point out that this model presents a high-level view, not so easily applicable to the operational level. In general, these frameworks demonstrate that there is a need to combine groups of quantitative results and qualitative results. The need for diversification in the PMS approaches, is also highlighted by Nudurupati, Tebboune, and Hardman (2016), who suggest the incorporation of “behavioural as well as environmental and social measures”, enlarging the collection of data to collaborative networks and social media. Beelaerts van Blokland et al. (2019) suggest what they call a third-generation business PMS or method that consists of evaluating intangible and non-financial dimensions, with indicators such as Conception, related to research and development, Configuration, related to Supply Chain, and Continuation, related to People Management, to thus obtain a "big picture" or "big story" about what happens within the organization and that helps to understand its complexity. Some of the proposals to advance PMS are pyramid-based models, which consider a hierarchy of organizational objectives including operational performance, and prism models based on a prism of performance evaluation perspectives including stakeholder satisfaction, strategies, processes, capabilities, and others (Perera and Perera, 2019; Ruano Pérez et al., 2018). For Ante et al. (2018), a structured framework of performance indicators is crucial for measuring the gap between the current state of operations and the desired state, and in many cases, it can be used to identify the path of progress in terms of overcoming productivity gaps. Therefore, they suggest a pyramid-based structure divided into three levels hierarchical levels: main system, sub-system and individual measures. The topmost level (main system) corresponds to the corporate vision, and financial and market objectives. At the intermediate level (sub-system), are the objectives regarding the maintenance of high productivity and quality, speed of response, flexibility, and lead times. And finally, the last level is related to the operations with indicators such as cycle time, loss of materials, number of failures, etc. As implicitly shown, all PMS are based on performance indicators related to the objectives of the organization and the relationship between those indicators. A performance indicator quantifies systems’ dimensions and behaviours, measuring how well an activity is being performed (Eckerson, 2009), and can act as an early warning sign that an unfavourable condition exists. A Key Performance Indicator (KPI) reflects the performance of an organization considering its key success factors. KPIs carry information relevant to managing operations at various levels in the organization. Through the measurement and continuous monitoring of KPIs, aspects in the production process are quantified and identified that allow the continuous improvement of the system (Kang et al., 2016b). According to Stricker, Echsler, and Lanza (2017), understanding the connection that a KPI has regarding the quantities measured in the system is important, but it is also important to understand the relationship between KPIs and the interdependencies inherent to them. Jooste and Botha (2018) argue that there is a limited understanding of the impact of KPIs on the projects or organization results and the impact of one KPI on another. KPIs have different directions, strengths, and polarities, i.e., the addition of some can cause the decrease of others, positive (the bigger the better), negative (the smaller the better). For Saiz, Bas, and Rodríguez (2007) when there is a deviation from a certain indicator, certain objectives are not achieved, making it hard for the manager to have early information on ALBERTO BUMBA ET AL. / PRODUCTION ENGINEERING ARCHIVES 2023, 29(2), 175-185 178 ARCHIWUM INŻYNIERII PRODUKCJI the causes of the problem, due to the lack of information associated with the deviations of the indicators, this is because a cause-and-effect relationship between the indicators is not established. The relationships between KPIs are established in several ways and using various techniques. Rodriguez, Saiz, and Bas (2009) use a method based on statistical data that consists of quantifying the cause-and-effect relationship between KPIs. They apply the Principal Component Analysis (PCA) method to determine the correlation coefficients. Jooste and Botha (2018) applied the same method improving it with Parallel Analysis (PA) and Screen Plot, for better identification of the indicators. Zhu et al. (2018) propose a structure for organizing KPIs. The structure is divided into process KPIs and measurement elements. Since process KPIs are dependent on the measuring elements, the latter are measured directly on the shop floor. According to Kang et al. (2016), statistical methods have an advantage in identifying the sign (positive or negative) of the relationships. However, they may not find intrinsic connections between indicators and their management ideas, in addition, the data collected in different production systems can lead to substantially different results. The same author proposes a multilevel hierarchical structure, which consists of three categories: supporting elements, intermediate KPIs, and comprehensive KPIs. Although the International Standard ISO 22400-1 (2014) and International Standard ISO 22400-2 (2014) report describes 34 KPIs, more stringent definitions are necessary to make clear the differences between them. It is necessary to redefine some KPIs and additional KPIs must be added. In addition, these KPIs must be classified logically, so that it is necessary to discover the intrinsic relationships between them. Therefore, it is necessary to group KPIs into various categories at various levels, which have explicit cross-links, and the KPI tree performs this function well. A Key Performance Indicator Tree (KPI tree) is a PMS in the form of a tree diagram which combines performance indicators in a hierarchical structure from the strategic objective at the highest level of the organization to operate at the lowest level, providing a transparent view of the status of all divisions of the organization at the strategic, tactical and operational level (Ante et al., 2018a). According to these authors, the KPI tree appears to respond to difficulties with the design of the entire structure of the PMSs, requiring the identification of appropriate key performance indicators (KPIs), the implementation of monitoring systems, and the identification of the relationship between the performance indicators. 3. Experimental This section presents the adopted research methodology and the initial characterization of the KPI structure in the studied company. 3.1. Methodology Case studies are used to clarify why a decision or set of decisions was made, how they were implemented and with what results were achieved. Moreover, a case study presents an analysis of a real-life problem, using practical methods to solve it and analyse its results (Easterby-Smith et al., 2018; Saleheen et al., 2014). As the main goal of this work was to improve the knowledge of the intrinsic relationships of KPIs in production systems, presenting details of a value stream PMS, based on a KPI tree, a case study approach was selected. Thus, a case study would allow to illustrate the application of the KPI tree in the industrial context, at the company where this concept of the KPI tree was also developed. Moreover, it would also allow studying the way a KPI tree would promote continuous improvement projects. The first step toward this study was to know how the KPI tree is built in terms of structure, indicator categories, and calculations. The second step was to know how the KPIs are linked to each other on the trees. After that, it would be possible to present some proposals for improvement of the utilization of the KPI tree in value stream continuous improvement projects. 3.2. KPI tree levels According to Ante et al. (2018), the KPI tree is a tool adapted to the current dynamics of organizations, it contemplates their critical success factors, and can be adapted and improved according to the needs of the organization. Such a tool may help in the implementation of continuous improvement projects, as it can indicate the focus of the actions. In addition, the levels of the KPI tree correspond to the levels of responsibility in the company, i.e., each entity in the company knows which part of the KPI tree to look to obtain the information they need. In this case study, the KPI tree is composed of five levels of indicators as presented by (Ante et al. 2018). • Level 1: Value Contribution It constitutes the existence of the business unit (factory target). Executive management is responsible for them. The value contribution KPI is the highest in the hierarchy, it gives a global view of what was delivered as value to the customer and what would come in a form of profit or contribution to the survival of the company when compared to the product selling price. In the context of value contribution are KPIs related to controlling, accounting and finance. For example, Planned Manufacturing costs are a controlling KPI which are divided into planned manufacturing variable costs and planned manufacturing fixed costs and so on. • Level 2: Key Performance Result (KPR) At the KPR level, there are performance indicators at a financial level, such as total cost, delivery services, and quality, which contribute to determining the overall value of a given value stream (product or area). Operation management is responsible for them. This level of KPI receives all the inputs from the level below and translates them into cost KPIs. This translation allows the controlling of consumption in terms of costs. For example, the number of end products in a warehouse, or the amount of work in progress can be translated into cost values on this level of KPIs. • Level 3: Value Stream KPR ALBERTO BUMBA ET AL. / PRODUCTION ENGINEERING ARCHIVES 2023, 29(2), 175-185 ARCHIWUM INŻYNIERII PRODUKCJI 179 At the Value Stream level, non-financial performance indicators for a given value chain are found, such as number of defects, productivity, and delivery performance. They constitute inputs for KPR. Value stream management is responsible for them. The value streams KPR are effective for value stream management, this level groups all the KPIs related to topics that assess the performance of the value stream in terms of quantities and time. They are supported by Monitoring KPIs, presenting a global and combined view of every product variant KPI. For example, by monitoring the produced quantities, the number of operators and working time, the productivity of product variants and as well as the productivity of the value stream can be obtained. • Level 4: Monitoring KPR At the Monitoring level, are found the indicators to execute and monitor the production system such as OEE, stock level, and Line Takt. They are inputs for Value Stream KPRs. Shop floor leadership is responsible for them. Monitoring KPIs are obtained by some sort of calculation that can be performed using standard formulas in line with the system features. Most of the KPIs of these levels are dependent on others to have some value or result. Kang et al. (2016) call these sets of KPIs Basic KPIs, according to them, basic KPIs reveal some performance aspect of work system, obtained from monitored data of supporting elements. They categorized the basic KPIs into three groups: production, quality, and maintenance. • Level 5: Improvement KPIs In terms of Improvement, there are indicators directly measured in the process, such as cycle time, defects, stops, and lack of resources. Improvement KPIs indicate a potential for improvement and areas of activity and constitute inputs for Monitoring KPIs. Operators are responsible for the Improvement KPIs. If these are well-defined, then obtaining improvement KPIs become a simple direct process to be implemented by the organization, either by collecting automatically with sensors or measuring manually. Improvement KPIs are considered elementary, so that performance can be identified and measured, making it easy to define the right actions for performance increase. (Kang et al. 2016) defines these as support elements, being divided into categories of time and quantity used to support improvement or basic KPI. 4. Results and discussion 4.1. Intrinsic relationship between Indicators The intrinsic relationship among indicators through the KPI tree follows a bottom-up approach, the improvements KPIs support the monitoring KPIs, and these are the KPRs. Improvements KPIs are used to collect data at the level of the production process on machines, operators, or even the combination of both in a single indicator. And it is from these data that the results of the above indicators are derived. These data indicators are essentially categorized as time or quantity (Kang et al., 2016b), which are subsequently worked with other factors to obtain the results of the elements above. The following is a demonstration of how KPI tree indicators linked to productivity are intrinsically related. A tree that can be used to report the productivity of a value stream, and all the indicators underlying it. It is important to note that trees such as the one below can exist for several KPRs related to quality, delivery performance, and indirect productivity. The demonstration is done from top to bottom, starting from KPR to the Improvements KPI that supports them. The indicators will be identified by (I) improvement, (M) Monitoring, and (R) KPR for better categorization according to the levels of location in the trees. Table 1 shows the units of the indicators represented in the KPI tree. Table 1. KPI tree productivity indicator units Unit Description Unit Description pcs/mhr parts per hour man min minutes Nr labors, unity event events h hour min/event minutes per event pcs parts pcs/event part per event sec Seconds pcs/sec part per seconds The formulas used to describe the relationships are based on the case study. At the top of the productivity KPI tree, Fig. 1, is the production line productivity (WPD): 𝑊𝑃𝐷 = 𝑂𝑢𝑡𝑝𝑢𝑡 𝑁𝑝𝑐𝑑∗𝑊𝑜𝑟𝑘𝑖𝑛𝑔ℎ𝑜𝑢𝑟𝑠 [𝑝𝑐𝑠 𝑚 ⁄ℎ𝑟] (R), - 𝑂𝑢𝑡𝑝𝑢𝑡[𝑝𝑐𝑠] - quantity of produced parts (M), - 𝑁𝑝𝑐𝑑[𝑁𝑟] - number of direct line employees (I), - 𝑊𝑜𝑟𝑘𝑖𝑛𝑔ℎ𝑜𝑢𝑟𝑠[ℎ] - observed production period (M). The number of employees is considered an Improvement KPI, and it can be measured directly through observation. As shown in Fig. 1, the productivity of the Value stream can be obtained through the relation of the productivity of the existing lines in it. Fig. 1. KPI tree Productivity for a Value Stream (Production Line) At the level of Monitoring KPI is the Output of the line, Fig. 2 shows its connection: 𝑂𝑢𝑡𝑝𝑢𝑡 = 𝑃𝑂𝑇 ∗𝑂𝐸𝐸[𝑝𝑐𝑠] (M), - 𝑃𝑂𝑇[𝑚𝑖𝑛] - planned production time (M), - [𝑝𝑠𝑐 𝑠𝑒𝑐 ⁄ ] - line cycle time (M) e - 𝑂𝐸𝐸[%] - effectiveness indicator (M). Fig. 2. Line Output Output Npcd Working hours Line x (Productivity Direct Employees) VS Productivity Direct Employees Line x (Productivity Direct Employees) Line x POT (Planned Operating Time) Output Line x LT (Line Takt) Line x OEE ALBERTO BUMBA ET AL. / PRODUCTION ENGINEERING ARCHIVES 2023, 29(2), 175-185 180 ARCHIWUM INŻYNIERII PRODUKCJI POT, Fig. 3, obtained by: 𝑃𝑂𝑇 = 𝑆ℎ𝑖𝑓𝑡𝑡𝑖𝑚𝑒 −𝐿𝑒𝑔𝑎𝑙𝑏𝑟𝑒𝑎𝑘𝑠 − 𝑃𝑙𝑎𝑛𝑛𝑒𝑑𝑠𝑡𝑜𝑝𝑝𝑎𝑔𝑒𝑠[𝑚𝑖𝑛], - 𝑆ℎ𝑖𝑓𝑡𝑡𝑖𝑚𝑒[𝑚𝑖𝑛] - shift duration time (I), - 𝐿𝑒𝑎𝑔𝑎𝑙𝑏𝑟𝑒𝑎𝑘𝑠[𝑚𝑖𝑛] - time for legal breaks and the (I) - 𝑃𝑙𝑎𝑛𝑛𝑒𝑑𝑠𝑡𝑜𝑝𝑝𝑎𝑔𝑒𝑠[𝑚𝑖𝑛] - planned downtime (M). Fig. 3. Planned operating time structure. 𝑆ℎ𝑖𝑓𝑡𝑡𝑖𝑚𝑒 and 𝐿𝑒𝑔𝑎𝑙𝑏𝑟𝑒𝑎𝑘𝑠 are predefined basic times. While the planned downtime, Fig. 5, is obtained by: 𝑃𝑙𝑎𝑛𝑛𝑒𝑑𝑠𝑜𝑝𝑝𝑎𝑔𝑒𝑠 = 𝑇𝑃𝑀 + 𝑆ℎ𝑖𝑓𝑡𝑐ℎ𝑎𝑛𝑔𝑒 + 𝑠𝑎𝑚𝑝𝑙𝑒𝑝𝑟𝑜𝑑𝑢𝑐𝑡𝑖𝑜𝑛 +𝑃𝑙𝑎𝑛𝑛𝑒𝑑𝑚𝑒𝑒𝑡𝑖𝑛𝑔𝑠 + 𝑃𝑙𝑎𝑛𝑛𝑒𝑑𝐶𝐼𝑃𝑎𝑐𝑡𝑖𝑣𝑖𝑡𝑖𝑒𝑠[𝑚𝑖𝑛], - 𝑇𝑃𝑀[𝑚𝑖𝑛] - duration of productive maintenance (M), - 𝑆ℎ𝑖𝑓𝑡𝑐ℎ𝑎𝑛𝑔𝑒[𝑚𝑖𝑛] - shifts change time (I), - 𝑆𝑎𝑚𝑝𝑙𝑒𝑝𝑟𝑜𝑑𝑢𝑐𝑡𝑖𝑜𝑛[𝑚𝑖𝑛] - sample production time (I), - 𝑃𝑙𝑎𝑛𝑛𝑒𝑑𝑚𝑒𝑒𝑡𝑖𝑛𝑔𝑠[𝑚𝑖𝑛] - planned meeting time (I), - 𝑃𝑙𝑎𝑛𝑛𝑒𝑑𝐶𝐼𝑃𝑎𝑐𝑡𝑖𝑣𝑖𝑡𝑖𝑒𝑠[𝑚𝑖𝑛] - time for continuous improvement activities (I). Among these KPIs, only the TPM, Fig. 5, is not an Improvement KPI, which is obtained by: 𝑇𝑃𝑀 = 𝐴𝑢𝑡𝑜𝑛𝑜𝑚𝑜𝑢𝑠𝑚𝑎𝑖𝑛𝑡𝑒𝑛𝑎𝑛𝑐𝑒 + 𝑃𝑙𝑎𝑛𝑛𝑒𝑑𝑚𝑎𝑖𝑛𝑡𝑒𝑛𝑎𝑛𝑐𝑒[𝑚𝑖𝑛], where the first part is autonomous maintenance time and the second is planned maintenance time, respectively. 𝐴𝑢𝑡𝑜𝑛𝑜𝑚𝑜𝑢𝑠𝑚𝑎𝑖𝑛𝑡𝑒𝑛𝑎𝑛𝑐𝑒𝑡𝑖𝑚𝑒, is obtained by: 𝐴𝑢𝑡𝑜𝑛𝑜𝑚𝑜𝑢𝑠𝑚𝑎𝑖𝑛𝑡𝑒𝑛𝑎𝑛𝑐𝑒 = ∑𝐷𝑢𝑟𝑎𝑡𝑖𝑜𝑛𝑜𝑓𝑠𝑖𝑛𝑔𝑙𝑒𝑡𝑎𝑠𝑘𝑖∗𝐹𝑟𝑒𝑞𝑢𝑒𝑛𝑐𝑦𝑖 𝑛 𝑖=1 [𝑚𝑖𝑛] (M) - 𝐷𝑢𝑟𝑎𝑡𝑖𝑜𝑛𝑜𝑓𝑎𝑠𝑖𝑛𝑔𝑙𝑒𝑡𝑎𝑠𝑘[𝑚𝑖𝑛] – activity duration (I), - 𝐹𝑟𝑒𝑞𝑢𝑒𝑛𝑐𝑦[𝑁𝑟] – number of times the activity is performed (I) e - 𝑛[𝑢𝑛] – Number of tasks to be performed (I). The planned maintenance time is obtained by: 𝑃𝑙𝑎𝑛𝑛𝑒𝑑𝑚𝑎𝑖𝑛𝑡𝑒𝑛𝑎𝑛𝑐𝑒 = ∑𝐷𝑢𝑟𝑎𝑡𝑖𝑜𝑛 𝑜𝑓 𝑠𝑖𝑛𝑔𝑙𝑒𝑡𝑎𝑠𝑘𝑖∗𝐹𝑟𝑒𝑞𝑢𝑒𝑛𝑐𝑦𝑖 𝑛 𝑖=1 [𝑚𝑖𝑛]. (M) 𝑂𝑢𝑡𝑝𝑢𝑡 it is also calculated with the cycle time, Fig. 4, and this one is obtained by: =∑𝐶𝑇𝑂𝑃𝑖 𝑛 𝑖=1 +𝐶𝑇𝑀𝐴𝐸𝑖[𝑠𝑒𝑐 𝑝𝑐𝑠 ⁄ ], - 𝐶𝑇𝑂𝑃𝑖[𝑠𝑒𝑐 𝑝𝑐𝑠 ⁄ ] - cycle time of an operation performed by the operator (I) and - 𝐶𝑇𝑀𝐴𝐸𝑖[𝑠𝑒𝑐 𝑝𝑐𝑠 ⁄ ] - cycle time of an operation performed by a machine (I). Fig. 4. Cycle time structure Cycle time structure 𝑂𝑢𝑡𝑝𝑢𝑡 is also calculated by 𝑂𝐸𝐸, Fig. 6, and this one is obtained by: 𝑂𝐸𝐸 = (1−𝑃𝑒𝑟𝑓𝑜𝑟𝑚𝑎𝑛𝑐𝑒𝑙𝑜𝑠𝑠𝑒𝑠 ∗ 𝑄𝑢𝑎𝑙𝑖𝑡𝑦𝑙𝑜𝑠𝑠𝑒𝑠 ∗𝐴𝑣𝑎𝑖𝑙𝑎𝑏𝑖𝑙𝑖𝑡𝑦𝑙𝑜𝑠𝑠𝑒𝑠)[%], - 𝑃𝑒𝑟𝑓𝑜𝑟𝑚𝑎𝑛𝑐𝑒𝑙𝑜𝑠𝑠𝑒𝑠[%] - line performance losses (M), - 𝑄𝑢𝑎𝑙𝑖𝑡𝑦𝑙𝑜𝑠𝑠𝑒𝑠[%] - loss of quality on the line (M) and - 𝐴𝑣𝑎𝑖𝑙𝑎𝑏𝑖𝑙𝑖𝑡𝑦𝑙𝑜𝑠𝑠𝑒𝑠[%] - availability losses (M). Fig. 5. Structure of planned stoppages Line x POT (Planned operating time) Line x TAZO (Breaks mandated by law) Line x TSCH (Shift time) Line x Planned stoppages CT MAE Line x LT (Line Takt) CT OP Line x Autonomous maintenance Line x Planned maintenance Line x TPM (Total Prod Maintenance) Line x Planned meetings Line x Shift change Line x Sample production Line x Planned,CIP activities Line x Planned stoppages Line x frequency Line x Duration of single task Line x frequency Line x Duration of single task ALBERTO BUMBA ET AL. / PRODUCTION ENGINEERING ARCHIVES 2023, 29(2), 175-185 ARCHIWUM INŻYNIERII PRODUKCJI 181 Fig. 6. OEE structure with loss of quality and availability The performance losses are obtained by: 𝑃𝑒𝑟𝑓𝑜𝑟𝑚𝑎𝑛𝑐𝑒𝑙𝑜𝑠𝑠𝑒𝑠 = 1− (𝑁𝑢𝑚𝑏𝑒𝑟𝑜𝑓𝑝𝑟𝑜𝑑𝑢𝑐𝑒𝑑𝑝𝑎𝑟𝑡𝑠∗𝑃𝑙𝑎𝑛𝑛𝐿𝑇 𝑁𝑒𝑡𝑜𝑝𝑒𝑟𝑎𝑡𝑖𝑜𝑛𝑡𝑖𝑚𝑒∗3600 )∗100[%], - 𝑁𝑢𝑚𝑏𝑒𝑟𝑜𝑓𝑝𝑟𝑜𝑑𝑢𝑐𝑒𝑑𝑝𝑎𝑟𝑡𝑠[𝑁𝑟] - quantity produced including good, defective and reworked (M), - 𝑃𝑙𝑎𝑛𝑛𝐿𝑇[𝑠𝑒𝑐 𝑝𝑐𝑠 ⁄ ] - planned cycle time (M) and - 𝑁𝑒𝑡𝑜𝑝𝑒𝑟𝑎𝑡𝑖𝑜𝑛𝑡𝑖𝑚𝑒[𝑚𝑖𝑛] - planned time of operation without losses due to availability (downtime, lack of material and employees) (I). The quality losses, Fig. 6, are obtained by: 𝑄𝑢𝑎𝑙𝑖𝑡𝑦𝑙𝑜𝑠𝑠𝑒𝑠 = (𝑁𝑂𝐾 ∗ 𝐿𝑖𝑛𝑒 𝑇𝑎𝑘𝑡 𝑃𝑂𝑇 )∗100[%], - 𝑁𝑂𝐾 - number of defective parts (I), - Line Takt - cycle time and (M) and 𝑃𝑂𝑇 - planned operating time (M). Being that, 𝑁𝑂𝐾 = 𝑅𝑒𝑗𝑒𝑐𝑡 +𝑅𝑒𝑤𝑜𝑟𝑘[𝑝𝑐𝑠], - 𝑅𝑒𝑗𝑒𝑐𝑡𝑠[𝑝𝑐𝑠] - quantity of rejected parts (I) and - 𝑅𝑒𝑤𝑜𝑟𝑘[𝑝𝑐𝑠] - number of reworked parts (I). The Availability Losses, Fig. 7, are obtained by: 𝐴𝑣𝑎𝑖𝑙𝑎𝑏𝑖𝑙𝑖𝑡𝑦𝑙𝑜𝑠𝑠𝑒𝑠 = (𝐶𝑂𝑙𝑜𝑠𝑠𝑒𝑠+𝑂𝑟𝑔𝑎𝑛𝑖𝑧𝑎𝑡𝑖𝑜𝑛𝑎𝑙𝑙𝑜𝑠𝑠𝑒𝑠+𝑇𝑒𝑐ℎ𝑖𝑛𝑖𝑐𝑎𝑙𝑙𝑜𝑠𝑠𝑒𝑠 𝑃𝑂𝑇 )∗100[%], - 𝐶𝑂𝑙𝑜𝑠𝑠𝑒𝑠[𝑚𝑖𝑛] - losses during changeover (M), - 𝑂𝑟𝑔𝑎𝑛𝑖𝑧𝑎𝑡𝑖𝑜𝑛𝑎𝑙𝑙𝑜𝑠𝑠𝑒𝑠[𝑚𝑖𝑛] - organizational losses (M) and - 𝑇𝑒𝑐ℎ𝑖𝑛𝑖𝑐𝑎𝑙𝑙𝑜𝑠𝑠𝑒𝑠[𝑚𝑖𝑛] - technical losses (M). Fig. 7. OEE availability structure Line x POT' Line x Availability Losses Line x Quality Losses Line x Line Tact' Line x Rework Line x Rejects OEE Line x Line x Performance Losses Line x Number of produced parts Line x Planned LT Line x Net operation time Line x NOK parts (not OK parts) OEE Line x Line x Availability Losses Line x Organizational Losses Line x Personnel missing Line x Raw material missing Line x Jam (due to next process) Line x Changeover Losses Line x Changeover Loss' Line x Changeover Time Line x Performance Loss c/o Line x Technical Losses (TA) Line x Number of changeovers Line x Line Tact'1 Line x Changeover time Internal Line x Output Line Line x Main part missing ALBERTO BUMBA ET AL. / PRODUCTION ENGINEERING ARCHIVES 2023, 29(2), 175-185 182 ARCHIWUM INŻYNIERII PRODUKCJI Changeover losses, Fig. 7, are obtained by: 𝐶𝑂𝑙𝑜𝑠𝑠𝑒𝑠 = 𝑁𝑢𝑚𝑏𝑒𝑟𝑜𝑓𝑐ℎ𝑎𝑛𝑔𝑒𝑜𝑣𝑒𝑟 ∗ 𝐶ℎ𝑎𝑛𝑔𝑒𝑜𝑣𝑒𝑟𝑙𝑜𝑠𝑠𝑒𝑠[𝑚𝑖𝑛], - 𝑁𝑢𝑚𝑏𝑒𝑟𝑜𝑓𝑐ℎ𝑎𝑛𝑔𝑒𝑜𝑣𝑒𝑟[𝑁𝑟] - number of changeovers made (I) and - 𝐶ℎ𝑎𝑛𝑔𝑒𝑜𝑣𝑒𝑟𝑙𝑜𝑠𝑠𝑒𝑠[𝑚𝑖𝑛] - changeover losses (M). The losses for each changeover, Fig. 7, are obtained by: 𝐶ℎ𝑎𝑛𝑔𝑒𝑜𝑣𝑒𝑟𝑙𝑜𝑠𝑠𝑒𝑠 = (𝑂𝑢𝑡𝑝𝑢𝑡𝐶𝑂 ∗)+ 𝐶ℎ𝑎𝑛𝑔𝑒𝑜𝑣𝑒𝑟𝑡𝑖𝑚𝑒𝑖𝑛𝑡𝑒𝑟𝑛𝑎𝑙[𝑚𝑖𝑛 𝑒𝑣𝑒𝑛𝑡 ⁄ ], - 𝑂𝑢𝑡𝑝𝑢𝑡𝐶𝑂[𝑝𝑐𝑠 𝑒𝑣𝑒𝑛𝑡 ⁄ ] - quantity produced during changeover (I), - [𝑝𝑐𝑠 𝑠𝑒𝑐 ⁄ ] – cycle time (M) and - 𝐶ℎ𝑎𝑛𝑔𝑒𝑜𝑣𝑒𝑟𝑡𝑖𝑚𝑒𝑖𝑛𝑡𝑒𝑟𝑛𝑎𝑙[𝑚𝑖𝑛 𝑒𝑣𝑒𝑛𝑡 ⁄ ] - internal changeover time (I). Organizational losses, Fig. 7, are obtained by: 𝑂𝑟𝑔𝑎𝑛𝑖𝑧𝑎𝑡𝑖𝑜𝑛𝑎𝑙𝑙𝑜𝑠𝑠𝑒𝑠 = 𝑃𝑒𝑟𝑠𝑜𝑛𝑒𝑙𝑚𝑖𝑠𝑠𝑖𝑔 + 𝑟𝑎𝑤𝑚𝑎𝑡𝑒𝑟𝑖𝑎𝑙𝑚𝑖𝑠𝑠𝑖𝑛𝑔 +𝑗𝑎𝑚[𝑚𝑖𝑛], - 𝑃𝑒𝑟𝑠𝑜𝑛𝑒𝑙𝑚𝑖𝑠𝑠𝑖𝑔[𝑚𝑖𝑛] - lack of collaborators (I), - 𝑟𝑎𝑤𝑚𝑎𝑡𝑒𝑟𝑖𝑎𝑙𝑚𝑖𝑠𝑠𝑖𝑛𝑔[𝑚𝑖𝑛] - lack of raw materials (I) and - 𝑗𝑎𝑚[𝑚𝑖𝑛] – obstruction during production (I). The technical losses, Fig. 8, are obtained by: 𝑇𝑒𝑐ℎ𝑖𝑛𝑖𝑐𝑎𝑙𝑙𝑜𝑠𝑠𝑒𝑠 = 𝐷𝑢𝑟𝑎𝑡𝑖𝑜𝑛𝑜𝑓𝑓𝑎𝑖𝑙𝑢𝑟𝑒 ∗ 𝑁𝑢𝑚𝑏𝑒𝑟𝑜𝑓𝑓𝑎𝑖𝑙𝑢𝑟𝑒[𝑚𝑖𝑛], - 𝐷𝑢𝑟𝑎𝑡𝑖𝑜𝑛𝑜𝑓𝑓𝑎𝑖𝑙𝑢𝑟𝑒[𝑚𝑖𝑛] - duration of the technical failure occurred (M) and - 𝑁𝑢𝑚𝑏𝑒𝑟𝑜𝑓𝑓𝑎𝑖𝑙𝑢𝑟𝑒[𝑁𝑟] – number of Occurrences (I). The duration of the failure, Fig. 8, is obtained by: 𝐷𝑢𝑟𝑎𝑡𝑖𝑜𝑛𝑜𝑓𝑓𝑎𝑖𝑙𝑢𝑟𝑒 = 𝑅𝑒𝑎𝑐𝑡𝑖𝑜𝑛𝑡𝑖𝑚𝑒𝑜𝑓𝑟𝑒𝑝𝑎𝑖𝑟𝑠𝑒𝑟𝑣𝑖𝑐𝑒 + 𝑅𝑒𝑎𝑐𝑡𝑖𝑜𝑛𝑡𝑖𝑚𝑒𝑜𝑓𝑜𝑝𝑒𝑟𝑎𝑡𝑜𝑟 +𝑇𝑖𝑚𝑒𝑟𝑒𝑝𝑎𝑖𝑟𝑡ℎ𝑒𝑓𝑎𝑖𝑙𝑢𝑟𝑒[𝑚𝑖𝑛] - 𝑅𝑒𝑎𝑐𝑡𝑖𝑜𝑛𝑡𝑖𝑚𝑒𝑜𝑓𝑟𝑒𝑝𝑎𝑖𝑟𝑠𝑒𝑟𝑣𝑖𝑐𝑒[𝑚𝑖𝑛] - repair service reaction time (I), Fig. 8. Structure of technical losses in the OEE - 𝑅𝑒𝑎𝑐𝑡𝑖𝑜𝑛𝑡𝑖𝑚𝑒𝑜𝑓𝑜𝑝𝑒𝑟𝑎𝑡𝑜𝑟[𝑚𝑖𝑛] - operator reaction time (I) and - 𝑇𝑖𝑚𝑒𝑟𝑒𝑝𝑎𝑖𝑟𝑡ℎ𝑒𝑓𝑎𝑖𝑙𝑢𝑟𝑒[𝑚𝑖𝑛] - repair time of failure (I). The observation period is part of productivity calculation, Fig. 9, and this is obtained by: 𝑊𝑜𝑟𝑘𝑖𝑛𝑔ℎ𝑜𝑢𝑟𝑠 = 𝑃𝑙𝑎𝑛𝑛𝑒𝑑𝑠𝑡𝑜𝑝𝑝𝑎𝑔𝑒𝑠 +𝑃𝑂𝑇[ℎ], - 𝑃𝑙𝑎𝑛𝑛𝑒𝑑𝑠𝑡𝑜𝑝𝑝𝑎𝑔𝑒𝑠[𝑚𝑖𝑛] - planned downtime (M), - 𝑃𝑂𝑇[𝑚𝑖𝑛] - planned production time (M). The planned production time in this branch, Fig. 9, is broken down as follows: 𝑃𝑂𝑇 = 𝑂𝑟𝑑𝑒𝑟𝑡𝑖𝑚𝑒 +𝐴𝑔𝑟𝑒𝑒𝑑𝐵𝑟𝑒𝑎𝑘𝑡𝑖𝑚𝑒 + 𝑆𝑐ℎ𝑒𝑑𝑢𝑙𝑒𝑑𝑚𝑎𝑖𝑛𝑡𝑒𝑛𝑎𝑛𝑐𝑒𝑡𝑖𝑚𝑒[𝑚𝑖𝑛], - 𝐴𝑔𝑟𝑒𝑒𝑑𝑏𝑟𝑒𝑎𝑘𝑡𝑖𝑚𝑒[𝑚𝑖𝑛] - break time (M), - 𝑂𝑟𝑑𝑒𝑟𝑡𝑖𝑚𝑒[𝑚𝑖𝑛] - time to order (M) (batch), - S𝑐ℎ𝑒𝑑𝑢𝑙𝑒𝑑𝑚𝑎𝑖𝑛𝑡𝑒𝑛𝑎𝑛𝑐𝑒𝑡𝑖𝑚𝑒[𝑚𝑖𝑛] - Planned maintenance time (M). The order time (batch), Fig. 9, is obtained by: 𝑂𝑟𝑑𝑒𝑟𝑡𝑖𝑚𝑒 = 𝐸𝑥𝑒𝑐𝑢𝑡𝑖𝑜𝑛𝑡𝑖𝑚𝑒 + 𝑆𝑒𝑡𝑢𝑝𝑡𝑖𝑚𝑒[𝑚𝑖𝑛], - 𝐸𝑥𝑒𝑐𝑢𝑡𝑖𝑜𝑛𝑡𝑖𝑚𝑒[𝑠𝑒𝑐] - execution time per unit (cycle time) (M), - 𝑆𝑒𝑡𝑢𝑝𝑡𝑖𝑚𝑒[𝑚𝑖𝑛] -preparation time (M). Fig. 9. Working hours KPI´s structure The execution time (time to produce an order), Fig. 10, is obtained by: 𝐸𝑥𝑒𝑐𝑢𝑡𝑖𝑜𝑛𝑡𝑖𝑚𝑒 = 𝐵𝑎𝑠𝑖𝑐𝑡𝑖𝑚𝑒 + 𝐴𝑙𝑙𝑜𝑤𝑎𝑛𝑐𝑒𝑡𝑖𝑚𝑒[𝑠𝑒𝑐], - 𝐵𝑎𝑠𝑖𝑐𝑡𝑖𝑚𝑒[𝑠𝑒𝑐] - basic time of an operation is obtained by (I), - 𝐴𝑙𝑙𝑜𝑤𝑎𝑛𝑐𝑒𝑡𝑖𝑚𝑒[𝑠𝑒𝑐] - time allowed for any interruptions (I). The basic time for an operation, Fig. 10, is obtained by: 𝐵𝑎𝑠𝑖𝑐𝑡𝑖𝑚𝑒 = 𝐴𝑡𝑖𝑣𝑖𝑡𝑦𝑡𝑖𝑚𝑒 +𝑊𝑎𝑖𝑡𝑖𝑛𝑔𝑡𝑖𝑚𝑒[𝑠𝑒𝑐], - 𝐴𝑐𝑡𝑖𝑣𝑖𝑡𝑦𝑡𝑖𝑚𝑒] – time to perform an operation (I), - 𝑊𝑎𝑖𝑡𝑖𝑛𝑔𝑡𝑖𝑚𝑒] - waiting time between operations (I). The activity time of an operation, Fig. 10, is obtained by: 𝐴𝑐𝑡𝑖𝑣𝑖𝑡𝑦𝑡𝑖𝑚𝑒 = 𝑖𝑛𝑓𝑙𝑢𝑒𝑛𝑐𝑖𝑎𝑏𝑙𝑒𝑡𝑖𝑚𝑒 + 𝑛𝑜𝑛𝑖𝑛𝑓𝑙𝑢𝑒𝑛𝑐𝑖𝑎𝑏𝑙𝑒], - 𝑖𝑛𝑓𝑙𝑢𝑒𝑛𝑐𝑖𝑎𝑏𝑙𝑒𝑡𝑖𝑚𝑒[𝑠𝑒𝑐] - time which operator impact product (I), - 𝑛𝑜𝑛𝑖𝑛𝑓𝑙𝑢𝑒𝑛𝑐𝑖𝑎𝑏𝑙𝑒𝑡𝑖𝑚𝑒[𝑠𝑒𝑐] - time which operator does not impact the product (I). The time allowed for any interruptions, Fig. 10, is obtained by: 𝐴𝑙𝑙𝑜𝑤𝑎𝑛𝑐𝑒𝑡𝑖𝑚𝑒 = 𝑇ℎ𝑒𝑐ℎ𝑖𝑛𝑖𝑐𝑎𝑙𝑡𝑖𝑚𝑒 +𝑃𝑒𝑟𝑠𝑜𝑛𝑎𝑙], - 𝑇ℎ𝑒𝑐ℎ𝑖𝑛𝑖𝑐𝑎𝑙[𝑠𝑒𝑐] - Interruptions due to technical situations (I), - 𝑃𝑒𝑟𝑠𝑜𝑛𝑎𝑙[𝑠𝑒𝑐] - Interruptions due to personal situations. Line x Time to repair the failure Line x Duration of failure Line x Technical Losses (TA) Line x Number of failures Line x Reaction time of repair Line x Reaction Time Operator Setup time Execution time Scheduled maint time POT Line x Working hours TWAF(Planned stoppages) Order time Agreed Break time ALBERTO BUMBA ET AL. / PRODUCTION ENGINEERING ARCHIVES 2023, 29(2), 175-185 ARCHIWUM INŻYNIERII PRODUKCJI 183 Fig. 10. Execution time, time to produce 10 units for example The union of the branches represented above results in a larger tree that covers almost all measured indicators that contribute to the productivity of a line, and consequently the value stream. The collection and availability of data to feed each indicator of the structure make the tool more robust and useful for system monitoring. The intrinsic relationship between the indicators helps in calculating the results of the indicators at the highest levels, with the values of the indicators at the lowest levels being directly measured. 4.2. KPI tree as a tool to support improvement projects Besides the use of the KPI tree to assess the performance of various indicators in the value stream, the company also uses it to define continuous improvement projects. However, this tool was not well used for this purpose, which led to a long time in the definition of projects, because there was no welldesigned structure like the KPI tree that could explain the root causes of the project and its impact on indicators at higher levels of the value stream. Thus, it was found that the definition time was 154 minutes per project, and often 10 to 30 projects are defined, which normally take a few days to define all. It was also found that only 38% of the projects had KPI trees, but these were not built according to the norms, which often made it difficult to discover the root causes of the projects, and consequently in the high execution time. The problems mentioned above had the following root causes: • No visual standard for KPI trees: There was no visual standard for the KPI trees used, which made it difficult to read and apply the created trees. • Inexistence of a standard to create a KPI tree: There was no standard tool to create the KPI tree, which made it difficult to implement in a project. • Difficulty in accessing the KPI tree documentation: Access to documents related to the KPI tree was difficult, leading people to a lack of knowledge about the KPI tree and consequently difficulty in using it. • No databases to feed KPIs: KPI tree data was not obtained automatically, it was placed manually. In addition, intrinsic relationships between KPIs were not established. The causes pointed out led to the lack of a structure of indicators that would allow a quick and relational analysis of them, to define projects in less time and, consequently, guarantee the presence of this tool during the definition, execution, and presentation of projects. To reduce the impact of these issues, the following proposals were developed: • Creation of a new visual standard: this solution allowed for better visual management of the indicators, allowing them to be differentiated based on their status, and making it easier for users to read them. • Development of a tool to create KPI trees: this solution allowed KPI trees to be created exclusively with a tool to ensure agility and maintenance of the visual standard created. • Organization and centralization of documents about the indicators and the KPI tree: this measure allowed easy access to documents about the KPI tree and its indicators. Which contributed to the increase of the knowledge of the people involved in the process. • Creation of a database and establishment of the intrinsic relations between the indicators: this measure allowed the obtainment of data and the calculation of the results of the indicators on the upper levels of the KPI tree in an automatic way. In this way, the tool to create a KPI tree has become more agile and effective for its proper use. The results obtained with the implementation of the previous proposals led to an increase in the number of projects with a KPI tree. During the project definition workshops, the tool was applied with functions that facilitated obtaining the data and reading the results indicators automatically and fast. Therefore, the percentage of projects with a KPI tree increased from 38% to 77% as shown in Table 2. Regarding to the time taken to define projects, it has also been reduced, since the phases of analysis of indicators that were previously carried out on sheets, and tables and took a long time, causing problems in reading the relationship between the indicators and updating their data. The problem Execution time Basic time Allowance time Personal Thechinical Activity time Waiting time Influenciable non influenciable