Capturing value through data-driven internal logistics: case studies on enhancing managerial capacity

ABSTRACT A key motivation to adopt data-driven solutions in facility-internal logistic systems is to create more responsive, efficient, and sustainable systems through a seamless data flow. There is thus a need to analyse the logistics system’s requirements from a data efficiency perspective. This paper presents and demonstrates a novel method, based on established literature, to assure that a transition to data-driven internal logistics leads to improvements in operational performance objectives. First, we identify wastes on the shopfloor caused by inefficient data flows. Second, we portray the trajectories of managerial capacities enhancements, supporting decision makers toward a to-be scenario. Two case studies are presented where the method has been implemented and used for demonstration and validation purposes. The results show that the method is beneficial in a waste-elimination and continuous improvement setting, linking improvements, detected wastes, enabling technologies, and managerial capacities in terms of monitoring, control, optimization, and autonomy.


Introduction
Facility-internal logistics processes involve the constant movement of goods, resources, and equipment on the shopfloor.These kinds of internal logistics (IL) systems are characterized by high dynamics and challenges in predictability, leading to a low level of visibility on the shopfloor and eventually creating wastes (Kalaiarasan et al., 2022).In accordance with the principles of lean manufacturing (Womack & Jones, 1996), this paper considers waste as any activity that does not contribute value to the final product or service.In general, this is analysed from value stream perspective where each stage needs to be investigated to identify various waste.Instances of waste in internal logistics encompass, but are not restricted to, excessive transportation, surplus inventory retention, inefficient search for items or information, and nonessential waiting times for goods arrival or departure.This situation in internal logistics is also an area of opportunity for smart connected technologies to create a seamless data flow, establishing an integrated IL system to mitigate waste creation (Parhi et al., 2022).Such data-driven solutions are expected to empower companies with more responsive, efficient, and sustainable intra-site logistic systems.The data-driven systems are described as creating a seamless data flow by linking data to the movement of goods and materials (Zhang et al., 2018).They are further associated with improved operational performance objectives (OPO) such as highly flexible mass production capability, real-time coordination, decentralized decision-making, improved quality, value chain optimization, and cost reduction (Moeuf et al., 2018;Zhang et al., 2018).
In order to meet the operational performance objectives (OPO), specific managerial and production capacities are mentioned as aligned with the concept of data-driven IL, the new smart connected technologies, and the internal logistics services (Moeuf et al., 2018).New set of functions and capabilities are enabled by intelligence and connectivity of the smart technologies; these are known as managerial capacities (Porter & Heppelmann, 2014).Managerial capacity represent monitoring, control, optimization, and autonomy within the warehouse where goods are received, put away, retrieved, palletized and delivered (Flores-García et al., 2022;Moeuf et al., 2018).Figure 1 illustrates how managerial capacities lead to operational performance objectives, enabled by technologies such as Internet of things (IoT) and cyber-physical systems (CPS) (Moeuf et al., 2018).
In forming this seamless data flow (Parhi et al., 2022), there is a need to have an efficient layout of the value stream processes from a data flow efficiency perspective (Redeker et al., 2019).This supports the need for translating data to value, and for matching the explorative activity of technology introduction to the exploitative potential in terms of managerial capacities (Colli et al., 2021).It has been concluded that the absence of an efficient data flow leads to waste at the operational level (Redeker et al., 2019).
Further, it has been argued that technologies have the potential to combat the seven types of wastes in lean manufacturing (Laaper & Kiefer, 2020;Pierluigi et al., 2019).
Hence, it is essential to spot the interruptions and wastes caused by inefficient data flow on the shopfloor.Addressing the managerial capacities without having a clear picture of the wastes caused by inefficient data flow is claimed to lead to complications in implementing data-driven approaches (Sharma et al., 2022).
Earlier research introduced methods to map information wastes on the shopfloor (Meudt et al., 2017;Roh et al., 2019).Their work focused primarily on detecting information stream wastes in administrative production-related systems, known as information logistics.Information logistics is defined in relation to IT administrative systems and aims to ensure optimized information provision and information flow (Filipe et al., 2009).However, there is no tool or structured approach for the evaluation of the information stream on the shopfloor to detect wastes caused by inefficient data flow (Yarbrough et al., 2022).Warehouse operations and IL operations benefit from systems such as warehouse management systems (WMS) or enterprise resource planning systems (ERP).Both these systems have the capacity to automate data collection, handle real-time data flow, and integrate data flow with the other sections of an enterprise.However, these tools have limited ability to analyse the shopfloor wastes in relation to the data flow in IL.In addition, it is challenging to explore the efficiency of the existing data flow structure in terms of meeting the shopfloor user requirements.In addition, there is a lack of empirical studies to evaluate how managerial capacities can be enhanced as a result of creating a seamless data flow in internal logistics.
In this research, we address this need for a method to create a seamless data flow as a basis for data-driven internal logistics, in turn enhancing key managerial capacities.To form this data flow and evaluate the enhancement of managerial capacities, the proposed method is built on two main pillars: the first concerns the detection of wastes that are caused by inefficient data flow, while the second addresses the relationship between identified process step improvements, detected wastes, enabling technologies, and managerial capacities.Hence, to develop this method, we have formulated two research questions: RQ1.How can data flow efficiency be assessed in relation to the shopfloor wastes in a current situation?RQ2.How can introducing data-driven technologies enhance managerial capacities and eventually reduce the detected wastes?
The novelty and contributions of this research are two-fold.Firstly, it provides managers and solution developers with practical guidelines for detecting wastes and areas for improvement in their data flow for internal logistics systems.These guidelines are based on established knowledge, but combined in a unique way.Secondly, it provides researchers with a method for studying the integration of enabling data-driven technologies into logistics operations and the related implications in terms of managerial capacities.The two empirical studies allowed us to perform an explorative test and demonstration of the proposed method, which led to a third contribution: an overview of the key challenges and contributions of a logistics re-design process.
The remainder of the paper is organized as follows.Section 2 presents the frame of reference, including the earlier works, concepts, and related definitions.The overall research approach and steps taken to develop and test the proposed method are presented in section 3. Section 4 is the presentation of the developed method.Section 5 presents the two case studies where the method has been implemented and used for demonstration and validation purposes.Section 6 presents the discussion and conclusion regarding the method, contributions, and future research possibilities.

Managerial capacities
In the literature, the managerial capacities that can be enhanced via data-driven approaches and a seamless data flow are divided into four main groups.Monitoring can be described as a real-time capability of products and machines to reveal their location and status (Ardolino et al., 2022).Control aims to control the system functions and personalize the user experience via controlling resources remotely through embedded software systems (Cimini et al., 2021).In some cases, there is a need to have man-machine interaction; thus, the level of control can be either semi-automated or fully automated (Moeuf et al., 2018).Optimization refers to the use of technologies, data models, and simulation systems to synchronize all the actors around the value chain (Mauricio-Moreno et al., 2015).Autonomy is a combination of monitoring, control, and optimization that can lead to autonomous problem-solving (Tao et al., 2018).Proper action against sudden changes in the logistics system, such as late or cancelled delivery from the suppliers, is an example of autonomy (Tao et al., 2018).Big data analytics (Zhong et al., 2017) Digital twin (Park et al., 2019) AI/Machine learning (Ding et al., 2020) Autonomy Resource planning (Moeuf et al., 2018) Delivery planning (Flores-García et al., 2022) Route planning (Flores-García et al., 2022) Cyber-physical system (CPS) (Hohmann & Posselt, 2019) Big data analytics (Zhong et al., 2017) AI/Machine learning (Ding et al., 2019) Table 1 gives an overview of the type of activities and examples of the enabling technologies for the respective managerial capacity based on the earlier studies by Moeuf et al. (2018), Flores-García et al. (2022), and Zafarzadeh et al. (2021).

Waste
Depending on the application, waste is defined from different perspectives.In this paper, we describe the literature on waste from two main points of view: waste on the shopfloor and waste in relation to information flow.

Waste on the shopfloor
In lean production, waste refers to any human activity that utilizes resources with no created value (Womack & Jones, 1996).In this context, there are 8 types of wastes known as Muda, shown in Table 2 (Lewis, 2000).
Lean production systems aim to minimize the waste and create business value by concentrating on customers' requirements (Womack & Jones, 1996).

Waste in relation to data flow
Considering the waste categorization in lean production, information management wastes have been described in relation to lean production wastes (Hicks, 2007;Romero et al., 2018).Four types of information management waste were identified, directly mapped to lean production wastes (Hicks, 2007): • Flow excess -time and resources that are needed to deal with excessive information i.e. information overload; flow excess equivalents-over-production • Flow demand -activities and resources that are needed to deal with a lack of information; this might involve either new information generation, or acquiring additional information; flow demand equivalents -waiting • failure demand -time and resources that are needed to identify the information; failure demand equivalents -over-processing Moving items and materials without adding value to the process or service.
Over-processing Refers to equipment and processes with low quality; overcomplicating a process is another aspect.

Excess inventory
Keeping inventories to cover shortages in the processes.

Defects
Can be both internal and external.Internal refers to scrap, rework, and delay; external refers to warranty, repair, and field service.

Unused talents
Poor human resource planning, where either the quality or the quantity of the resource allocation is inappropriate.
• Flawed flow -time and resources that are needed to correct or verify information; flawed flow equivalents -defects.
The fundamental causes of waste were categorized into four main groups by Hicks (2007): • Existence of the data -the data is not generated, or the process is broken, or a critical process is unavailable.
• Data presentation -the data flow is interrupted because the data presentation mode makes the data flow unable to be shared or difficult to be collected.• Timeliness -data generation happens too often or too late, so that the amount of generated data impacts the quality of the data.
• Data accuracy -inaccurate data flow leads to inappropriate downstream activities, corrective actions, or verification.
In a similar approach, wastes in lean information have been categorized to clarify nonvalue-added activities in administrative and information management systems (Filipe et al., 2009;Redeker et al., 2019).

Waste detection methods
With respect to waste in information flows, researchers have described methods to identify waste in different types of systems, such as on the shopfloor or in the administration systems.
On the shopfloor, the value stream includes all materials and information as well as their flow through the production system, including internal logistics (Chen et al., 2010).Even though the value stream mapping (VSM) method has been widely used in different industries, it contains shortcomings in relation to information stream mapping on the shopfloor (Roh et al., 2019).The survey by Lugert et al. (2018) showed that the provision of real-time performance indicators is one of the main items that is missing in relation to data sharing in value stream mapping.
Considering Industry 4.0, methods such as VSM require conceptual modifications to meet the digitalization requirements (Lugert et al., 2018).Concerning the need for a more advanced method for value stream mapping in the Industry 4.0 era, Meudt et al. (2017) suggested the method VSM4.0.The aim is to analyze the information logistics on the shopfloor based on traditional value stream mapping but targeting information logistics wastes.
Lean thinking can be applied to information management (Salvadorinho et al., 2021).The application of lean thinking to information management can enhance business values (Hicks, 2007).A value flow model applied to information management was proposed by Hicks (2007) and is shown in Figure 2. In this value-flow model, products flow to meet the demand of the customer, user, or consumer (another system), while information flows to add value to either information or work processes.Examples are data generation for decision-making support or historical analysis.Embracing the proposed value flow model supports waste elimination as it has been defined in section 2.2.However, no detailed explanation is provided for waste detection in this proposal.
The information stream mapping method enables an analysis similar to VSM, but it is specifically designed for information streams in a production environment (Roh et al., 2019).In this method, the concept of waste is in line with lean information, as explained in 2.2.2.The proposed method improves information streams by mapping and analyzing their characteristics, and a sequence diagram is used to map the individual transfer of information between different participants in the information stream.

Smart data-driven internal logistics framework for enhancing managerial capacities
Once the wastes have been detected, there is a need to drive improvement efforts towards operational performance objectives (OPO), as stated in the introduction, by focusing on managerial capacities.Several frameworks are presented in the literature that transform collected data into information and knowledge.According to the purpose of this study which is stated in the introduction, the to-be scenario should recognize the role of data to capture value and facilitate the enhancement of the managerial capacities.Thus, we identified frameworks based on two main criteria: frameworks that take data-driven logistics services into account, and frameworks that consider managerial capacities.The main reason to choose these criteria is their contribution to portraying the to-be scenarios.
One key framework that from a big-data perspective supports smart manufacturing, including logistic activities, is presented by Tao et al. (2018).According to the authors, a data-driven system has the following characteristics that imply its autonomy: • Self-regulation -by exploiting real-time monitoring of production processes.This will enable the system to have swift responses to unexpected events and failures.An example could be delayed raw material delivery from suppliers or machine breakdowns.
• Self-execution -by exploiting multi-source data from different manufacturing processes to rigorously control the production processes.For example, it will be possible to deal with material delivery dynamics that happen on the shopfloor.Material can be delivered to different production locations depending on the requirements.
• Self-organization -through exploiting resource-related data, tasks, and work instructions data to have smart planning and scheduling.Optimal resource allocation helps manufacturing systems to address manufacturing requirements in an efficient way.
• Customer-centric -by using data collected from different sources, it will be possible to offer customized products and services to customers.Customer demands, preferences, limitations, and behaviors are some examples of data that need to be exploited in a data-driven system.• Self-learning and self-adapting -by exploiting historical and real-time data to perform quality control and preventative maintenance proactively.As a result, quality defects and machine failures can be predicted and prevented before occurrence.
The framework consists of four modules as shown in Figure 3. information and actionable information.
(3) The real-time monitoring module receives data which have been transmitted through the manufacturing module and enables realtime monitoring of manufacturing.The ultimate goal of this module is to ensure quality despite changes and unexpected events.( 4) The problem-processing module enables the detection of problems, suggesting feasible actions, root cause analysis, and analysis of the potential impact on other parts of the manufacturing system.Real-time information and historical data help decision makers and intelligent systems to address both current and upcoming issues.
It should be noted that the characteristics of this proposed framework such as selflearning, self-organizing, and self-execution imply autonomy.

Gaps in the current methods and requirements of a future one
The literature reviewed in the preceding sections 2.1-2.4 corroborates the research gap identified previously.As stated in section 2.3 VSM does not have the capacity to support real-time information flow.This may lead to late response to the identified issues (Lugert et al., 2018).The earlier work by Meudt et al. (2017) is an initial effort to map the information flow and aims to reduce waste, but it lacks detail on how the information stream is acquired.In addition, it is not clear what information is required to be collected for each step of the process, and thus it becomes difficult to judge how to deal with the waste at that specific step.Furthermore, the proposed method does not give any indication about how to address the wastes indicated in the context of Industry 4.0.The work by Roh et al. (2019) focused on information stream mapping at a high level rather than focusing on issues at the shopfloor level.
Earlier works have used case studies in manufacturing as validation proof.However, it is not clear if these methods are applicable to internal logistics.Additionally, these methods aimed to reduce wastes in data flows.Thus, it is difficult to evaluate how business value can be gained through combating data wastes.Moreover, there is a need to indicate shopfloor wastes, as described in lean production, that have fundamental causes in lean information, as discussed by Hicks (2007).In conclusion, by analyzing the gaps in the existing methods and frameworks, the following requirements of a future method are identified: • A method should be able to be characterized as easy to learn and applicable to IL, supporting managerial capacities, and be business-value oriented.
• A method should be capable of detecting wastes on the shopfloor that are caused by data flow inefficiencies, known here as lean information wastes.• A method should be able to support the enhancement of managerial capacities via a waste reduction approach.

Research methodology
The main goal of this study is to develop and demonstrate a method for enhancing managerial capacities based on data flow efficiency in internal logistics settings.The approach and steps taken to meet the goal of this study are presented as follows.

Five moves for method development
As stated in section 1, the introduction of technology in internal logistics (IL) should fulfill the exploitative needs, aiming to enhance managerial capacities (Colli et al., 2021).Thus, it is important to identify the requirements of the IL systems from a data flow efficiency perspective.One of the main steps for assessing data flow efficiency is to elicit the problems and wastes that are caused by data flow interruptions on the shopfloor (Yarbrough et al., 2022).However, as stated in the introduction, there is a two-fold gap.First, there is no tool to assess data flow efficiency in relation to waste mitigation on the logistics shopfloor.Second, there is a need to have a framework that aligns the efforts in technology introduction to the identified wastes and eventually enhances managerial capacities.Figure 4 depicts the moves taken to cover this gap.
In Move 1, in order to clarify the issue, it was important to frame it from four different perspectives.First, we reviewed the literature to understand the managerial capacities, their respective activities, and enabling technologies (section 2.1).Second, it was important to scrutinize the literature on how to clarify the concept of waste both on the logistics shopfloor and from an information flow perspective (section 2.2).This also clarifies the dynamics between information flow and shopfloor operation.Third, based on the types of wastes described, we also identified previously described waste detection methods (section 2.3).Fourth and last, we reviewed the literature for frameworks with certain characteristics: targeting data-driven systems, including managerial capacities, and being applicable to internal logistics (section 2.4).Such previously described frameworks support in detail a method for streamlining the technology introduction efforts, mitigating the wastes, and portraying the enhancement trajectories.In move 2 we condensed the literature-based results from move 1, revealing the characteristics and requirements of the desired method.This is presented in section 2.5.In move 3, based on the characteristics and requirements from move 2, a structured method was proposed (section 4).It is important to note that, to create the proposed method, we synthesized elements of existing tools and definitions regarding information waste detection.Move 4 aimed to ensure the applicability of the proposed method, which was done by demonstrating it in two real IL environments.We made a comparison between two single case studies.The criteria to choose these companies are described in section 3.2.Finally, in move 5 we analyzed and discussed this approach of literature-based method development and then demonstrated the applicability of the method through case studies.

Case studies: case selection and data collection
The case study method facilitated an in-depth understanding of the potential benefits and challenges of a transition toward data-driven production logistics (Yin, 2018).In general, case study has shortages to generalize the findings.However it is considered as a suitable research method when the amount of unknown knowledge is considerable (Meredith, 1998).As mentioned earlier, the number of empirical studies regarding digitalisation of internal logistics systems are limited and in-depth knowledge in this area is scarce.Yin (2018) stated that case study is a suitable choice when (1) the researcher has little or no control over the course of events; (2) the focus of the study is contemporary.Hence, we decided to perform two in-depth study in two single cases.The main criterion for choosing the companies was to have two different cases with the same unit of analysis but different characteristics within internal logistics.The unit of analysis is the internal logistics flow, which starts from the receiving of goods and finishes with delivery to the production lines.'Case A' has a long history of heavy automotive manufacturing and assembly.'Case A' has the vision to be a frontier within a 'smart factory' and Industry 4.0.Consequently, the company is interested in streamlining its supply chain processes, including internal logistics.The internal logistics flow volume is very high and it involves many complex processes.'Case B' is active in the courier industry and has a noticeable amount of deliveries on a daily basis.Similar to the earlier case, 'Case B' is actively improving its internal logistics processes, aiming to assure high quality of deliveries as well as streamlining the processes.
In general, a case study protocol was developed to ensure validity and reliability of the study.Data collection followed the triangulation principle to increase the validity of the findings by minimizing the potential biases, and errors associated with any single data source or method (Kvale, 2007).The data collection sources included observation, document review, and semi-structured interviews.The collected data during the site visits were used to map the current state and later compared with the data obtained from the documents review and interviews.The current state were mapped by following data: digitalization strategies, working processes, KPIs, workshop layouts, IT systems, data flows, production data, and production infrastructures.The results of the as-is mapping, data flow mapping, shortages with the data flow, and proposed to-be scenario were presented to both companies in several occasions.The construct validity was further strengthened by peer examination of the collected data and the results derived from the case studies at different times by project members, industrial partners and research fellows.Building primarily explanations based on the analysis of the empirical data and literature helped to support internal validity in the case studies.Replication logic is used to assure external validity (Yin, 2018).The interview questions were derived from the research questions in order to ensure consistency among the case studies.To avoid bias, the results of the interviews were checked with another expert and reviewed with each interviewee prior to any further development within the research process.The semistructured interviews allowed the researcher to ask follow-up questions and ensure correct interpretation of the interviewees' statements.
We actively collected, documented, and analyzed the data.Each of the companies was visited in the presence of a host who had detailed knowledge regarding the working processes.In 'case A', an internal logistics manager, a digitalization development engineer, an internal logistics manager, an internal logistics expert, and an IT digitalization expert (IT super user) were interviewed.In 'case B' two IT digitalization experts (IT super users), a terminal manager, and two internal logistics experts were interviewed.The interviews aimed to collect information regarding IL work processes, IL performance monitoring methods, existing IT systems supporting IL operation, current data flow, issues with the existing data flow, and IL digitalization strategies and plans.The aforementioned data were collected through semistructured interviews to assure a precise understanding of the studied phenomenon (Kvale, 2007).During the site visits, the IL work process steps were double-checked with the operators.In addition, the IL experts were asked to describe the effectiveness of the existing data flow based on appendix A. In mapping the current situation, both cases provided us with the documents describing the IL processes, data maps, shopfloor layouts, and digitalization strategy.To assure quality and avoid misinterpretation, the study results were later presented and discussed with the interviewees in a separate meeting.

Proposed method
To cover the abovementioned gaps, a method with two phases is proposed, as depicted in Figure 5. Phase I identifies the shopfloor wastes caused by inefficient data flow.Phase II portrays the managerial capacities enhancement trajectories in a to-be scenario.The datadriven framework by Tao et al. (2018), presented in section 2.4, is used to organize the enhancement trajectories in terms of monitoring, control, optimization, and autonomy.

Phase I: waste detection
This phase focuses on waste detection on the shopfloor.It is similar to value stream mapping and aims to capture enhancement opportunities for internal logistics processes.In this phase, the wastes on the shopfloor are identified using a waste walk approach, similar to VSM.The mapping information needs to be collected by means of interviewing the system users, including the operators.To perform the interviews, a guideline can be used which is based on lean information adopted from Roh et al. (2019).The guiding questions are presented in Appendix A.
In order to describe each process, and find the fundamental causes of waste in each step of the IL process, seven terms need to be detailed.These are defined according to the lean information concept and are listed to the left in Figure 6.These terms and their respective description and examples are presented in detail in Appendix B.
In Figure 6, the process steps are written at the bottom.For each step, the required or output data is written on the first two rows.The source of the required data should be visualized through an arrow and a box representing the data producer or data consumer.In addition, information regarding the data presentation mode, data collection interval, and data collection mode for the corresponding process step needs to be filled in the rows shown in Figure 6.The last step is to note down the type of wastes that occurs at each step.
To be able to fill in the layout shown in Figure 6, an illustration of the detailed steps of Phase I is proposed in Figure 7.After mapping the IL work process flow and identifying the IL work process steps, either the required data or the output data to perform the task need to be identified.By using the guiding questions from Appendix A, any shortages or lack of data need to be marked.In the next step, the terms in Appendix B need to be mapped.The aim is to note down any shortages that might be related to the data presentation mode.In turn, the timeliness of the data needs to be considered to see if the current data timeliness meets the IL system requirements as shown in Figure 7.The questions are used to evaluate the answers for each step.In cases where there is an IT system or equipment involved at that step, the user should be interviewed.
Next, the data collection method used by the operator or the system user needs to be investigated.We suggest triangulating the data sources to assure data accuracy.In the end, all the shortages that were marked earlier need to be consolidated to identify the type of waste that occurs at each working step.

Phase II: managerial capacities and to-be scenario
In this second phase, the managerial capacities that were described in section 2.1 are used to map the waste reduction trajectory.
Based on the results from phase l, the process steps and respective wastes are documented as in Figure 8.By considering the types of activities described in Table 1, each of the process steps can be mapped in relation to at least one of the managerial capacities as well as the enabling technologies.For example, parts identification is considered as a monitoring capacity.Thus, the introduction of enabling technologies has the potential to reduce the wastes that happen in parts identification and enhance the monitoring capacity.After this step, the data-driven framework proposed by Tao et al. ( 2018) (section 3.4) should be used to determine the enhancement trajectories in the form of monitoring, control, optimization, and autonomy.The ultimate goal is to reduce the wastes by introducing enabling technologies and eventually enhancing managerial capacities.By portraying the to-be scenario, a holistic picture of the enhancement trajectories will be obtained.This will create a roadmap for the decision makers in order to streamline the efforts for transition towards a data-driven IL system.

Demonstration of the proposed method in two case studies
Two single case studies are used to demonstrate and verify the applicability of the proposed approach to identifying waste in relation to inefficient data flow, as well as suggesting enhancement trajectories.

Case A: a study of a logistics center at an automotive manufacturer
Case A is a European manufacturer of heavy commercial vehicles such as trucks, buses, and engines.The company operates in over 100 countries and has around 50,000 employees worldwide.In 2020, the company delivered 66,900 trucks and 5,700 buses.The logistics center (LC) is responsible for delivering parts to 'case A' internal customers.Every day almost 2,000 pallets and boxes are delivered to the LC by 'case A' suppliers.
As depicted in Figure 9, after the incoming trailers are registered, the pallets and boxes are unloaded by a lift truck and the parts are registered in the ERP system.Some of the pallets and boxes receive a new tag indicating the storage address.The tags are necessary for sorting the pallets and boxes based on their destination.The lift truck drivers need to read the tags to find the addresses.The sorted pallets are moved to the assembly facility.The internal customer registers the orders in the ERP system between 24 and 8 hours before the production.The orders are printed out and manually sorted.After this step, a picker is called to pick the parts and deliver them to the quality control (QC) unit to check if the correct part has been picked.Each order is equivalent to one pick.The controlled parts are consolidated and moved to a trailer, which departs the assembly facility four times per hour.

Phase I: detecting wastes caused by poor data flow in case A
Operator's motion, searching, waiting, unused talents, and over-processing are the types of wastes identified due to data flow inefficiency in case A, as shown in Figure 9.To  investigate the waste causes, the data collection method, data generation interval, and data visualization method were investigated, as shown in the rows in Figure 9.For instance, 'order sorting' involves an extensive search for data and rearrangement for the operator.The arrival of the pallets registers in the ERP system via scanning.Although the process is semi-automated, the arrival time might differ from reality.This is because the operator might scan the pallets at any time during the working shift.The lack of realtime data regarding the arrival time of the parts forces the logistics center to keep the parts available on the shelf to avoid parts shortage.
The act of 'pallet sorting' requires the operator's movement and searching to read the storage address.This is because the storage address and the part id are printed on a label attached to each of the pallets.Although the required information exists, the mode of presentation creates extra movement for the operator.The logistics center has no real confirmation of the shipped items.This makes it difficult to assess the service quality.In the 'order picking', the operator has to search for the information on the picking list to control compliance with the picked parts.The consolidated parts are checked by the QC team to assure each customer will receive the ordered parts together and not in separate turns.For resource planning, the logistics manager or the planner needs to estimate the trucks' arrival time and the quantity of the incoming items.In addition, information regarding lift truck availability and human resources needs to be manually collected by the manager.Sometimes the planning activity needs to be iterated in order to adjust to the changes.All these time-consuming activities require optimization to combat the wastes.

Phase II: data-driven to-be scenario in case A
In this phase, the managerial capacities for respective IL activity are mapped.As shown in Figure 10, each process step, the corresponding identified waste, and the managerial capacity to address the wastes are presented for 'case A'.
To mitigate the identified waste, some of the enabling technologies mentioned in Table 1 are proposed in Figure 11. Figure 11 suggests a possibility to link the four managerial capacities as well as the enabling technologies to mitigate waste and portray the enhancement trajectories in 'case A'.Applicable technologies are auto-ID technologies, RTLS, Pick by X technologies, and smart devices such as tablets and mobile phones.As there already are valuable data in the current ERP and WMS systems, these data need to be consolidated to facilitate real-time monitoring of the system, as shown in the data driver module in Figure 11.
There is the possibility to use auto-identification means such as RFID, QR codes, NFC tags, etc., or image processing to identify and register the incoming goods.As a result, manual registration will be omitted from the process.The order-sorting process is handled manually based on the chassis ID and destination address.This process can benefit from data analysis technologies to sort the orders, with no need for physical printing of the labels.The orders can be sorted by using machine-learning techniques and the result can be communicated to the pickers in real time.
The pickers need to collect the labels manually and then pick the parts, and later the picked parts are checked by QC personnel.There are several possibilities to use Pick by X technologies such as smart glasses, pick by voice, pick by light, etc., as shown in Figure 11.The pickers' location and operational status are unknown to the managers as no tracing is possible.Using real-time tracing technologies such as RTLS can facilitate the scheduling and planning processes.
The same situation prevails for the delivery trailers, as their location and operational status are unknown to the LC managers.After delivering the parts to the assembly lines, there is no procedure to confirm the delivery.
The LC managers assume that the shipped parts have been delivered to the right spot at the right time.In case of deviation, mistakes are reported by assembly line personnel to LC.At present, there is no real-time communication among the logistics units and systems.There is a possibility to use smart devices alongside cloud services to interconnect all the systems, as real-time data sharing will have a major role in the future plan.
In conclusion, the monitoring module constantly performs state analysis and realtime monitoring as shown in Figure 11.In this case, the inventory level, pickers' location, delivery status, resource availability, and material ordering are the items to be monitored to reduce the identified wastes on the shopfloor.In case of a shortage in the number of resources, the problem-processing module proposes a solution to the decision makers.The introduction of material handling automation technologies such as AMR and pick by X enhances the control aspect of the managerial capacities.Wastes in resource planning and order sorting can be addressed through optimization and the introduction of enabling technologies such as data analytics and machine learning.The autonomy of the system is enhanced by introducing CPS and AI in combination with other technologies to make the system self-adjusting, support the decision makers, and eliminate the identified wastes such as inefficient resource estimation.

Case B: a terminal at a courier company
Case B offers communication and logistics solutions to businesses and individuals.The company operates in the Nordic and Baltic regions, and has around 29,000 employees.Case B delivers mail and parcels to households and businesses, as well as providing e-commerce and logistics solutions to businesses.
The incoming trailers arrive at the terminal according to a fixed schedule, which is visible to the terminal.Customers' addresses are written on signboards hung on each port.This means the ports are static and cannot be used for any other customer or destination.The process steps and the corresponding wastes are identified in Figure 12.The trailers dock in to the incoming ports and one person is responsible for unloading the trailer with a lift truck.As the contents of the trailers are unknown at the terminal, there might sometimes be a need to have an extra resource to complete the unloading.Then the pallets are dispatched based on the final delivery address.To load the pallets to the trailers, an operator has to rearrange the pallets.The pallets are finally shipped to the customers based on the orders registered on the EDI (electronic data interchange) system.

Phase I: detecting wastes caused by poor data flow in case B
The study shows that all the required data for pallet unloading, lining up, and loading, as well as planning, need to be collected manually, as reflected in Figure 12.This involves wastes such as operator's motion, searching, inventory, waiting, and over-processing.The information regarding inbound truck arrival time is accessible through the 'truck fleet system'-see Figure 12.However, there is no information about the quantity, destination, and dimension of the incoming pallets.This lack of information forces the planners to estimate the truckload.Unknown delivery addresses of the incoming pallets in combination with the rigid line arrangement in the terminal lead to uneven distribution of the pallets on the shopfloor.In some cases, there is not enough space for delivery to one customer, so the operators have to temporarily keep the pallets on other lines.The situation becomes worse when the incoming pallets have to remain overnight.In such a situation, the operators have to keep the pallets inside a trailer next to the terminal due to space shortage, which leads to over-processing.
It should be noted that the delivery orders are registered in the EDI system.However, this data is not visible for the daily operations of the other stakeholders in the terminal.On the other hand, sometimes the EDI registered order differs from the actual order in terms of quantity, date, or destination.To unload and scan the incoming pallets, the operator has to read the delivery address manually, which involves motion.To line up the delivery, it was found that each of the outgoing gates was dedicated to one to three customers.The name of the customer is written on a board, which makes the lineup process rigid and inflexible.
The operator has to rearrange the line of pallets several times based on the final delivery destination.The ordered line of pallets helps the driver to unload the pallets without the need for reloading the trailer.

Phase II: data-driven to-be scenario in case B
Each process step, the identified wastes, and the respective managerial capacity are shown in Figure 13.For example, the types of wastes identified in unloading and registration need to be addressed via monitoring capacity.
Figure 14 suggests a possibility to link the four managerial capacities as well as the enabling technologies to mitigate waste and portray the enhancement trajectories in case B. To mitigate the identified waste, some of the enabling technologies described in section 3 are proposed in Figure 14.In the data driver module, data needs to be collected from different sources such as smart devices, auto-ID technologies, vision systems, RTLS, geo-fencing, and port displays.These data need to be integrated and pre-processed to create meaningful patterns and the results analyzed for visualization or further use.The processed data can be stored for either historical analysis or predictive analysis.The realtime monitoring module increases visibility by constant monitoring of operational performance.In this case, the lift trucks' movements need to be monitored in order to detect any deviation from the plan.RTLS technology is one possibility for real-time monitoring of the internal traffic.Resources need to be allocated to the right task at the right time.For example, the number of resources that are needed to unload an incoming trailer can be monitored.In case of a shortage in the number of resources, the problem-processing module will introduce a solution by considering the available resources.The port allocation will be dynamic to increase floor-space utilization efficiency.Therefore, it is important to monitor the port allocation process in real time in order to avoid any misplacement of pallets.
The customer information, departure time, number of pallets, etc. can be visualized on digital displays dedicated to each of the outgoing ports.In addition, the operator responsible for loading a trailer can constantly update the central system with smart devices such as tablets or mobile phones.Vision systems can control the loading process to double-check if the pallets being loaded are the correct ones.To omit the sorting process, an AI system can sort the order by placing pallets in front of each of the ports and communicating the results to a responsible operator through a tablet or smartphone.As a result, when the operator unloads an incoming truck, the auto-ID, such as RFID, will recognize the pallet.The AI system has access to all the other pallets that are supposed to be shipped with this pallet.As a result, it is possible to identify the pallet position in the loading queue.
Similar to case A, the monitoring module constantly performs state analysis and realtime monitoring, as shown in Figure 13.In this case, the internal transportation, port allocation, resource utilization, and shopfloor space allocation are the items to be monitored to reduce the identified wastes on the shopfloor.In case of deviations in inbound or outbound transportation, the problem-processing module proposes a solution to the decision makers.The introduction of material handling automation technologies such as AMR and pick by X enhances the control aspect of managerial capacities.Wastes in resource planning and port allocation can be addressed through optimization and the introduction of enabling technologies such as data analytics and machine learning.The autonomy of the system is enhanced by introducing CPS and AI in combination with other technologies to make the system self-adjusting, support the decision makers, and eliminate the identified wastes such as inefficient resource estimation.

Discussion and conclusion
Based on the literature review, the research questions are formed around two issues.The first is assessing the current situation regarding inefficient data flow effects on the shopfloor, while the second is identifying the managerial capacities enhancement trajectories.The following discusses these matters in more detail.

Assessing the current state of IL systems
It is argued that wastes have vicious cycles and that one form of waste may lead to another (Romero et al., 2019).In this context, we realized that there is no tool or framework to identify IL shopfloor wastes caused by inefficient data flow.To cover this issue we have proposed a method to detect and document such wastes.This is beneficial since it elucidates the expected values of transition to a data-driven IL system.In addition, the proposed approach helps to match the explorative aspect of technology introduction to the exploitative needs.Some of the earlier works (Hicks, 2007;Meudt et al., 2017;Roh et al., 2019) attempted to adapt the concept of waste in lean to the context of data flows.Thus, they discussed wastes in information flow, such as stocking excess data with no actual use.However, they did not address the relationship between shopfloor waste and data flow inefficiency.In this work, we have benefited from the results of these earlier works to identify the fundamental causes of information wastes.We synthesized elements from value stream mapping tools with tools to identify information wastes into a novel method aimed at mapping the shopfloor wastes that have originated from the data flow inefficiencies.To investigate the fundamental cause of the identified data flow inefficiencies we used guiding interview questions adopted from Roh et al. (2019).It is essential now to gather different viewpoints for further analysis.

Enhancing the managerial capacities
Digitalization of IL activities without addressing shopfloor wastes leads to complications in the transition phase (Romero et al., 2018).To address this issue we sought to portray the enhancement possibilities of managerial capacities based on a data-driven framework suggested by Tao et al. (2018).This framework facilitates the establishment of an efficient data flow by eliminating the shopfloor wastes that are caused by data flow inefficiencies.Proposing this framework fills the identified gap regarding the lack of a method to support managerial capacities enhancement through a waste reduction approach, as discussed in 2.5.
In this framework, the real-time monitoring module is associated with monitoring as one of the managerial capacities.In the proposed method, the monitoring involves real-time connection with the shopfloor tools, equipment, products, and sensors.Controlling and optimization of the processes are linked with the data driver module.Within this module, any deviation from the standards and routines can be noted and adjusted accordingly.The real-time data in addition to the historical data facilitate the optimization of the processes.The overall structure of the framework aims to make the IL system self-regulating, self-learning, selfexecuting, and self-organizing.
Demonstration of the proposed approach in two companies revealed that each of the managerial capacities could be enhanced in the following manner.

Monitoring by real-time capability
Many of the in-house logistics process steps require some level of monitoring in the following forms: • Real-time identification of the items across the IL flow through auto-identification technologies; real-time positioning of the items across the IL flow through RTLS technologies; • The possibility to trace back the events through auto-identification and sensor networks; • Real-time control of the inventory level through vision systems, sensor networks, and auto-id technologies; • The possibility to have real-time monitoring of environmental factors such as temperature, humidity, shock, etc. through sensor networks, auto-identification, etc.

Controlling by means of deviation management
Controlling the execution of the plans according to the predefined methods and standards is the application of control items for the case companies.Examples are controlling delivery from storage in accordance with the plan and pallet arrangement for loading to the trailers.The managerial benefit is the possibility to optimize material flow through interconnecting transportation equipment such as forklifts, AGVs, and delivery trains.

Optimization via resource management
The companies that were studied suffer from a lack of input data for the optimization of their daily operations such as resource planning, routing, and scheduling.However, the proposed approach in this paper revealed the following managerial benefits: • Possibility to optimize picking routes through interconnecting pickers' locations in real time for case A. • Possibility to optimize resource planning via implementing a data-driven approach.
Possibility to optimize layout planning via having flexible port allocation in case B. • Possibility to optimize the sorting order to create the picking sequence by considering the picking route, delivery priority, and delivery address.

Autonomy via decentralized decision making
Proper action against sudden changes in the logistics system, such as late or canceled delivery, is an example where the case companies can benefit from data-driven logistics.For example, road conditions and their impact on delivery time can be addressed via technologies such as AI.In general, the number of cases for potential use is less apparent, but the autonomy of the system lies in its capability to adapt and adjust its behavior according to events and incidents.

Overall method conclusions
This paper has proposed and demonstrated a novel method to identify wastes caused by inefficient data flow in internal logistics flow.The method provides a holistic map of the existing data flow on the shopfloor in order to organize the identification of enhancement opportunities.The method helps IL decision makers in order to streamline the efforts for a transition towards a data-driven IL system.The in-house logistics of two companies were studied for demonstration purposes.The results show that the proposed method is applicable to identifying shopfloor wastes caused by inefficient data flow known as lean information wastes.Additionally, the method helped to recognize the enhancement trajectories in the form of monitoring, control, optimization, and autonomy.Furthermore, the proposed method is based on well-established elements from standard procedures, while being combined, adapted and applied to fit internal logistic systems.The well-established foundation of the method also ensures ease of implementation and use in industry.

Limitations and future research
The unit of analysis in the cases used for demonstration is limited to a part of the IL flow -from receiving goods, to delivery, to production, or outbound.Hence, the applicability of the proposed approach cannot be generalized to other IL areas.The proposed approach has been demonstrated in two companies.However, having a greater number of cases will be helpful for evaluating the pros and cons of the proposed method.
In addition, both companies are considered to be large enterprises.The applicability of the proposed method can also be evaluated for SMEs.The purpose of this method is to identify and document information wastes in internal logistics settings as a basis for improvement toward a seamless data flow and data-driven internal logistics, in turn enhancing key managerial capacities.This method is intended to be used in a waste-elimination and continuous improvement setting, linking improvements, detected wastes, enabling technologies, and managerial capacities.However, it is evident that in an investment and implementation situation, a full cost/benefit analysis is needed, whereas this method is limited to the initial possibility identification phase.
For future research, the proposed approach needs to be fully validated in situations with higher and lower levels of automation.Moreover, the case studies were limited to the IL flow from receiving goods to delivery.Other processes, such as material delivery in the manufacturing line, can help validate the model in more detail.There is also the possibility of merging information logistics with the proposed method, aiming to draw a holistic picture of information flow and value stream mapping.

Figure 2 .
Figure 2. Value-flow model as applied to information management by Hicks (2007).

Figure 4 .
Figure 4.The research moves for the composition, demonstration, and analysis of the proposed method.

Figure 5 .
Figure 5. Overview of the proposed method.

Figure 6 .
Figure 6.Documenting the process mapping and waste detection, Phase I.

Figure 7 .
Figure 7.The process flow of waste detection in phase I.

Figure 8 .
Figure 8. Documenting the link between waste, managerial capacity, and enabling technology, Phase II.

Figure 9 .
Figure 9. Value and data flow analysis of the logistics center in case A.

Figure 10 .
Figure 10.Layout of the mapped managerial capacities in relation to the identified wastes case A.

Figure 11 .
Figure 11.To-be data-driven structure for case A.

Figure 12 .
Figure 12.Value and data flow analysis of the terminal in case B.

Figure 13 .
Figure 13.Layout of the mapped managerial capacities in relation to the identified wastes case B.

Figure 14 .
Figure 14.To-be data-driven structure for case B.

Table 1 .
Relationship between the managerial capacities, IL activities, and respective enabling technologies.

Table 2 .
Waste definition based on lean production.
Enterprise systems: any IT systems that already exist within the organization and have an application in relation to internal logistics, e.g.smart autonomous technologies such as IoT that require no human intervention to generate and visualize data, enterprise resource planning (ERP) system, warehouse management system (WMS), electronic data interchange (EDI), etc., e-communication, including email, chat systems, etc. Verbal communication: including phone calls, direct verbal communication, etc. Physical: including letters, papers, etc.Real-time data: generating data in parallel with an event.On-demand: generating or collecting data only when the user asks for them.Conditional: data are created at certain intervals regardless of the user needs, for example, every week, or once a change occurs.Automated: data are collected systematically with no need for human intervention, such as machine-tomachine communication (M2M) or real-time visualization for the end user.Semi-automated: data exist on IT systems such as ERP systems; however, data need to be visualized via human actions such as barcode scanning.Manual: human resources need to collect data manually, including observation or estimation.To evaluate the type of wastes caused by earlier stages.Wastes include the eight typical types of waste in lean production.