IN
DEGREE PROJECT MECHANICAL ENGINEERING,
SECOND CYCLE, 30 CREDITS ,
STOCKHOLM SWEDEN 2020
Configuration of an electronic
Kanban board for planning
analysis activities at an industrial
laboratory
ELIN ANDERSSON
JOHANNA NUGIN
KTH ROYAL INSTITUTE OF TECHNOLOGY
Configuration of an electronic Kanban board for planning
analysis activities at an industrial laboratory
Elin Andersson
Johanna Nugin
KTH Industrial Engineering and Management
Production Engineering
Abstract
To improve productivity, it is important that the right product is produced in the right quantity when that product is needed. Kanban is a way to signalise the need for material and its main aspect is that it pulls materials through the processes only when needed. Kanban can therefore help reduce Work In Progress (WIP).
Quality Control (QC) at AstraZeneca in Södertälje is tasked with controlling the quality of batch samples from production. Several different analyses need to be performed on a batch and all the analyses should preferably be done at the same time to reduce the levels of ongoing work. The currently used system and working procedures for planning analysis activities at the QC laboratories results in high levels of WIP. The aim of this project was to investigate the possibility to replace the currently used planning system with an electronic Kanban board so that WIP can be limited and a more predictable flow achieved. Observations and discussions have been conducted to understand the current state of the QC process. A literature review has been conducted to form a theoretical framework for concepts such as Just-In-Time and Kanban. A Kanban board has been configured using a software suggested by AstraZeneca for one team.
The functionality of the board is presented along with a discussion of how it can be used in their
operations. It has been configured so it is possible to prioritise the completion of ongoing batch analyses before new batches are started, making it possible to reduce WIP and achieve a more predictable flow. Tasks are to be prioritised according to a decision tree. Limitations in the construction of the board and challenges with a potential future implementation are discussed. The board has been presented to team leaders and analysts to gain their input. Based on this, some alterations have been suggested. A suggestion for the setup of a pilot test is presented.
It is recommended that the board is tested in parallel with the current planning system to evaluate the possibility to implement the board in a larger scale and to further investigate challenges and possibilities with the software.
Sammanfattning
För att förbättra produktiviteten är det viktigt att rätt produkt produceras i rätt mängd när den behövs. Kanban är ett sätt att signalera behov av material och dess huvudaspekt är att material dras genom processen endast när det behövs. Kanban kan därför minska pågående arbete (WIP).
Quality Control (QC) på AstraZeneca i Södertälje har som uppgift att kontrollera kvaliteten på
batchprover från produktion. Flera olika analyser måste genomföras på en batch och alla analyser ska helst slutföras samtidigt för att reducera nivåerna av pågående arbete. Det nuvarande systemet och arbetsrutiner för att planera analysaktiviteter på QC laboratorierna resulterar i höga nivåer av WIP.
Syftet med detta projekt var att undersöka möjligheten att ersätta det nuvarande planeringssystemet med en elektronisk Kanbantavla så att nivåerna av WIP kan begränsas och ett mer förutsägbart flöde uppnås. Observationer och diskussioner har genomförts för att få en förståelse för nuvarande arbetssätt i QC. En litteraturstudie har genomförts för att bilda ett teoretisk ramverk för koncept som Just-In-Time och Kanban. En Kanbantavla har konfigurerats för ett av lagen i en programvara som föreslagits av AstraZeneca.
Tavlans funktionalitet presenteras och det diskuteras hur denna kan användas i deras verksamhet. Den har konfigurerats så att det är möjligt att prioritera att analyserna av pågående batcher slutförs innan nya batcher påbörjas, vilket möjliggör en reduktion av WIP och skapandet av ett mer förutsägbart flöde. Ett beslutsträd ska användas som hjälpmedel för att prioritera arbetsuppgifter rätt. Begränsningar i
konstruktionen och utmaningar med en potentiell framtida implementation diskuteras. Tavlan har presenterats för team leaders och analytiker för att erhålla deras input. Baserat på denna har också några ändringar föreslagits. Ett förslag på upplägg för en pilottestning presenteras.
Det är rekommenderat att tavlan testas parallellt med nuvarande planeringssystem för att utvärdera möjligheten att implementera den i större skala samt för att vidare undersöka utmaningar och möjligheter med programvaran.
Acknowledgements
This report is the result of a master thesis project at the Institute of Industrial Production at The Royal Institute of Technology in Stockholm. This project has been conducted at AstraZeneca in Södertälje during the spring of 2020.
We want to thank our supervisor Mats Bejhem at the Royal Institute of Technology, for contributing with valuable insight, support and guidance.
We also want to thank our supervisor Peter Aurosell at AstraZeneca for having us, providing guidance and help with connections within the organisation.
Furthermore, we want to thank all the employees at AstraZeneca that have taken time to show us around and answer questions.
Stockholm, June 2020
Abbreviations
AZ- AstraZeneca
AZSS- AstraZeneca Supply System GMP- Good Manufacturing Practice JIT- Just-In-Time
KM- Control Method (Kontrollmetod) SHE- Safety Health Environment TL- Team Leader
QA- Quality Assurance QC- Quality Control WIP- Work in progress
Contents
1. Introduction ... 1
1.1 Background ... 1
1.2 Aim and objectives ... 2
1.3 Research questions ... 2
1.4 Delimitations ... 2
1.5 Methodology... 2
2. Theoretical framework ... 5
2.1 Lean Production ... 5
2.2 Toyota Production System (TPS) ... 5
2.3 7+1 wastes ... 5 2.4 PDCA cycle ... 6 2.5 Just-In-Time (JIT) ... 6 2.6 Pull vs. Push ... 7 2.7 Kanban ... 8 2.8 Visual management ... 11
3. Current situation at QC AstraZeneca Sweden Operations ... 13
3.1 Lean at AstraZeneca... 13
3.2 The process for a sample ... 14
3.3 Current planning activities for samples in QC... 15
4. Configuration of an electronic Kanban planning system ... 19
4.1 Introducing a pull system ... 19
4.2 Description of the software Kanbanize ... 20
4.3 Constructing a Kanban board at QC AstraZeneca Sweden Operations ... 24
5. Pilot testing of board ... 34
5.1 Initial pilot testing ... 34
5.2 Next phase of pilot testing ... 35
6. Discussion and conclusion ... 37
6.1 Limiting WIP and introducing a pull system ... 37
6.2 Challenges in work environment ... 37
6.3 How the limitations with E-plan are eased ... 38
6.4 Kanban board configuration ... 38
6.5 Future work and further suggestions of improvement ... 40
Appendices ... 1
Appendix A- Board construction and functionality ... 1
Appendix B- Business rule construction ... 4
1
1. Introduction
The following section provides an introduction to the thesis, where aim and objectives, research questions, delimitations and methodology are introduced.
1.1 Background
Industrial companies have a constant need to improve productivity in order to maintain and increase their competitiveness. In order to improve productivity, it is important to work towards ensuring that the right quantities and items are produced at the right time without waste, the Just-In-Time (JIT) concept. Kanban, originally a Japanese term, is a way to signal and visualise the need for materials in the
manufacturing industry. It is a concept within the production philosophy JIT. A Kanban system visualises what to produce, how much to produce and when to produce. The main aspect of a Kanban system is that it pulls material through the processes only when it is needed. If material is only refilled when needed, overproduction can be avoided. In this way, a Kanban system can help reduce waste by minimising the Work In Progress (WIP), unevenness in production and overloading for machines and operators (Oakland, 2014).
AstraZeneca (AZ) is a global and science-led pharmaceutical company delivering medicines worldwide. In the year 1999 the two companies Astra AB in Sweden and Zeneca Group PLC in United Kingdom merged into AstraZeneca (AstraZeneca, 2018a). Their focus is on oncology, cardiovascular, metabolic and respiratory diseases. AstraZeneca pursue research in areas of autoimmunity, neuroscience and infection. There are 70 600 co-workers in the world and AstraZeneca operates in over 100 countries, with research centres located in Sweden, USA and United Kingdom and production in 16 countries (AstraZeneca, 2020a). One of the world’s largest production and supply units of pharmaceuticals, Sweden Operations, is located in Södertälje. 13 billion tablets and capsules are manufactured yearly in Södertälje. The
manufactured volumes stand for around 35% of the total volumes sold by AstraZeneca (AstraZeneca, 2020b). The pharmaceutical industry is regulated by the principles of good manufacturing practice (GMP). The purpose of GMP is to ensure safe products for patients. GMP regulates, for example, the
documentation and recording of events, controlling and traceability of materials, maintenance of equipment and the control of processes (Sandle, 2015).
Quality Control (QC) at Sweden Operations is tasked with controlling the quality of products and has around 380 employees. QC ensures solid products for its patients and is divided into five different sections (AstraZeneca, 2019). QC is responsible for testing and analysing samples at multiple points through the receiving, formulation and packing process. Batch samples from production are analysed according to different control methods and the results are then evaluated and reviewed before the batch can be released by Quality Assurance (QA). It is an essential part in the supply chain as the results from QC are required for products to be released to the customers.
To plan the analysis activities, different planning systems are used within the QC organisation. E-plan is a non GMP planning system currently being used in several laboratories within Sweden Operations to plan analysis activities and to monitor the status of the analyses as well as the outcome. It is an electronic planning tool that consists of several linked Excel files. The use of this planning system and their current ways of working results in high levels of Work In Progress (WIP) because there is no restriction in how many batches can be in progress at the same time. The analysts are able to start analysing batches independently from one another.
2
1.2 Aim and objectives
To achieve a more predictable flow of work by minimising the Work In Progress (WIP), the planning of analysis activities needs to be optimised by introducing a pull system. The goal is to achieve this by configuring an electronic Kanban system. Instead of using the elements that are typical within the manufacturing industry, terms that are typical for industrial laboratories are used- sample queue,
preparations, analysis and checks. A review of the possibilities to use a commercially available software for the implementation of an electronic Kanban system is conducted. The advantages and disadvantages of the configured Kanban system are discussed, and suggestions for further work is presented.
The aim of this project is to present a suggestion for a configuration of a non GMP Kanban system for Quality Control (QC). Due to the extent and spread of QCs operations, the system needs to be
electronically visualisable.
1.3 Research questions
The aim of this project can be summarised into the following main research question:
● What are the expected advantages and disadvantages with an implementation of the suggested electronic Kanban system?
To answer this question, following sub-questions have been developed:
● How can the limitations with the current use of E-plan be eased by implementing an electronic Kanban system?
● What can the expected difficulties be with implementing a new system?
1.4 Delimitations
Focus will be on a commercially available software that has been suggested by AstraZeneca. Focus will not be on how the software is constructed, but rather how it can be used to configure a Kanban board for the stated purpose. A Kanban board will be constructed using data from only one of the teams. The constructed board will be presented to only a few of the team’s analysts who will be given the opportunity to test the configured board for a few articles.
1.5 Methodology
The method of this master thesis project consists of a literature study conducted simultaneously with field studies. These are then combined into a report and a suggestion for configuration of an electronic Kanban system, see figure 1 for the entire process.
The literature review has been conducted to form a theoretical framework for relevant concepts. These concepts include JIT, pull systems, Kanban, Visual Management, 7+1 wastes and other relevant concepts in the area of Lean. The literature review was conducted using databases such as DiVA, KTH Primo and Google Scholar as well as books. The researched sources have been analysed and critically reviewed. The information from research articles and books have been compared to ensure the correctness of the
3
Kanbanize is the software that has been reviewed for suitability in regard to a potential future implementation of an electronic Kanban system. In order to understand and learn the functions in
Kanbanize, the published tutorials have been studied as well as their public Knowledge Base (Kanbanize, 2020a) which contains articles about the different functionalities and possibilities that the software offers. To achieve an understanding of the current status, the currently used planning system E-plan has also been studied. This has been done in order to understand how it works and to be able to determine what
information and which functionalities are necessary for a transformation to an electronic Kanban system. E-plan is also used to retrieve necessary data. This is data about which control methods that should be performed and by which team.
To better understand the needs of a transformation to a Kanban system and the current conditions, field studies have been conducted at two different teams, Oncology/Multi (MU1) and Metoprolol/Candesartan (MU2). This has been done to gain better knowledge of how the analysts plan their day and to help identify which factors and conditions affect this planning. To understand what current conditions in the laboratories limit and affect the planning of the analyses, the different steps in the QC process have been observed. The field study has consisted of seeing how the samples are entered into E-plan in the sample receiving, how the analyst retrieves the samples and how the analyst performs different analyses on the samples. It has also consisted of seeing how an analyst plans his activities in the current system E-plan, as well as how the team commonly plan its activities in E-plan during the daily planning meetings.
A board with data from one team has been constructed in the software. This board has been presented to team leaders and analysts to discover and discuss potential drawbacks and problems that need addressing before an actual implementation is feasible in a larger scale. From the received input some alterations have been made to the board and future work is suggested.
4
Figure 1: The process of this master thesis project where observations along with a literature review have been conducted to configure a Kanban board.
5
2. Theoretical framework
In this section, the relevant literature for this project is studied and presented in order to answer the research questions. The section begins with a brief description of Lean and its origin. Concepts relevant for this project, such as JIT, pull systems and Kanban are then further described.
2.1 Lean Production
Lean production is a philosophy based on principles deriving from Toyota Production System (TPS), which is a well-known concept used today in the industry, but also in healthcare. Lean is a strategy which can be used for handling resources and making operations become more effective. It has to be integrated into the company’s core values and the strategy should be implemented differently in different
organisations. The main goal is to eliminate waste and create value for the customer. All activities at the company are categorised by whether they create value or not and the activities that do not add value are eliminated by using different tools and techniques. Lean philosophy refers to the whole organisation and by stepwise implementing standardised methods it will be ensured that everyone is working after the same methods. Visualisation of the methods makes it easier to remember, understand and improve. The Lean principles’ most important task is to provide guidelines and lead the company in the right direction. A central part of the Lean philosophy is continuous improvements which can be achieved by Just-In-Time (JIT) and Jidoka. An easy method to pass on the Lean principles is to make a model of Lean, which has the shape of a house (Petersson, Johansson, Broman, Blücher, & Alsterman, 2010).
2.2 Toyota Production System (TPS)
The Toyota Production System was developed by the founder of the Japanese company Toyota Motor Corporation. The main idea was to eliminate waste (muda). The costs can be minimised within the production by eliminating unnecessary work and inventory, which can be achieved by constantly improving all the activities in the company. TPS consists of 14 principles called the Toyota way which can be divided into; long term philosophy, the right process will produce the right results, add value to the organisation by developing your people and that continuously solving root problems drives organisational learning (Petersson et.al., 2010).
2.3 7+1 wastes
In order to achieve the most efficient operations it is vital to plan in a way that allows waste to be observed and simply removed. This should be made with controlled methods so the organisation can manage the improvements, by handling the aberrations and ensure that they will no longer occur. It is vital to be able to separate what work is waste and what work is value adding, therefore all the main focus must lay on decreasing everything that does not add value. The seven plus one wastes presented below are a key part of Lean and categorises waste in eight main types (Petersson, et.al, 2010).
1. Overproduction 2. Waiting 3. Transport 4. Inappropriate processing 5. Inventory 6. Motion
7. Producing defective products
6
2.4 PDCA cycle
The PDCA cycle is a method used in order to continuously develop and aim for improvement. It can be applied in many different areas for instance in production. The cycle begins at the normal situation in the production and when a problem or deviation occurs a plan is made in order to handle it. The plan is then tested and evaluated. Adjustments are applied and the cycle as shown in figure 2 can start over. PDCA stands for Plan, Do, Check, Act (Atanelov, 2016).
● Plan: Identify the problems and develop a plan ● Do: The plan from the previous step is executed
● Check: The results are gathered, evaluated and compared ● Act: Adjustments are made in order to meet the desired result
Figure 2: Shows the PDCA cycle and its four phases.
The Plan phase is about defining the needs of the customer, gathering data and identifying the origin of the problem. During this phase communication is very important, since everyone concerned in the project needs to be involved in order to get the best result.
The Do phase is where the plan is implemented and tested.
The Check phase includes checking if the work has been conducted as planned and evaluating results of the test. Evaluating the results includes discussing what worked well or did not work as well as if there are any changes that need to be done and what can be improved for future work.
In the last phase Act, new work methods are established and standardised to confirm that the work will be performed properly. The last phase of the PDCA cycle will be complete when the new methods have been documented. The cycle will restart where new improvements can be made (Roos, 1992).
2.5 Just-In-Time (JIT)
One of the fundamental principles in Lean is the Just-In-Time (JIT) concept. JIT requires that the organisation strives to deliver the right product, in the right amount, at the right time and without any waste. The right amount is decided by customer demand. Control of time is essential to achieve JIT and minimise waiting times and buffers. JIT can be described as continuous flow production (Oakland, 2014). There are three essential principles belonging to JIT: takt, continuous flow and pull system (Petersson et.al., 2010).
Takt
The takt describes the production rate (volume/time unit) and derives from the demand. The takt levels the rate of production. To be able to plan production and coordinate tasks, it is important that the production is
7
levelled. By levelling, it means to plan the production so that the level of work is constant over time. The takt time is the time which is dedicated to the individual product in every process and can be visualised on a takt board where it is possible to see the current status. The takt time can change over time (Ledbetter, 2018). Cycle time is the rate at which it is possible to produce. When the takt time equals the cycle time, you are producing according to JIT.
One essential part of takt is to detect deviations. When a part is not finished within the takt time there is waste in the flow. Takt makes it possible to detect even small deviations that otherwise may have gone undetected. Having daily control meetings makes it possible to compare the day’s outcome with the takt. This makes it possible for instant actions to be taken if deviations have occurred (Petersson et.al., 2010).
Continuous flow
The design of the production flow has a vital influence on the continuous flow. Products should constantly be in motion, as every stop leads to waiting time which is a form of waste. The more waiting time, the longer the lead time (the time between initiation and completion of a task). A continuous flow can be achieved by minimising stops, shorten distances between different stations and implement buffers between stations. In order to achieve a continuous flow, it is important to handle the conditions that are causing a need for inventory and Work In Progress (WIP). This can be caused by long change-over times which often lead to increased batch sizes. Large batch sizes will increase the need for buffering. To achieve a continuous flow the strive should be a batch size of one, a so-called one-piece flow (Petersson et.al., 2010).
Another important factor when establishing a continuous flow is the culture of the team. The workers need to be multi-skilled and it also requires the improvement of teamwork. Communication and information sharing is important as the workers are dependent on each other (Ledbetter, 2018).
Pull system
A pull system is controlled by demand and the production is triggered by a demand signal in a process. The information flow and the production flow are therefore running parallel. To avoid overproduction, unnecessary buffers, transportations and waiting times, nothing should be produced unless it is demanded. Having an established flow is a precondition to be able to implement a pull system. To achieve a pull system, a standardised method is essential in order to visualise a demand further down in the supply chain. A Kanban system is an easy technique which can be used to send signals between different processes. A levelled production is also required in order to create a stable flow (Petersson et.al., 2010).
One of the benefits with introducing a pull system is that the WIP levels decrease as the processes are controlled. Limiting the WIP levels are also beneficial for decreasing lead times. By visualising the WIP levels in a pull system it is also possible to discover where in the processes problems occur, thus aiding problem-solving (Ledbetter, 2018).
2.6 Pull vs. Push
The differences of a pull and push system is the process which triggers the drive of work. A push system decides the work schedule depending on the information coming from outside the system. A pull system is releasing work depending on information given from inside the system. The pull system decides how much raw material and parts are released, which means that a limitation for the maximum level of
8
inventory will exist. By using cards or signals, the system can be controlled, but in a push system the release of parts is not controlled because it is determined beforehand (Khojasteh, 2016).
With a pull system the inventory levels will decrease, and costs be reduced, also transportation and waiting time will be reduced. JIT is a known pull system used in order to control the inventory. The objective is to keep the inventory as low as possible and have just enough inventory to meet customer demand (Petersson et.al., 2010).
2.7 Kanban
Kanban is an essential part of the Just-In-Time production system, which has the goal of reducing and ultimately eliminating all types of wastes. It enables parts to be produced when they are demanded and in the quantity in which they are demanded. In this way production flow and inventory levels can be
controlled (Sendil Kumar & Panneerselvam, 2007). The main goals of Kanban are to limit WIP, visualise the workflow and measure the lead time. Kanban can be used as an aiding tool to work more effectively by visualising what should to be done. It uses Kanban cards as a signal to trigger tasks. These cards are moved between different stages of the processes as the work progresses (McLean & Canham, 2018). A pull system such as Kanban offers many advantages such as reduced inventory levels, shorter lead times, better error detection, less variability, increased productivity and improved quality (Khojasteh, 2016). Kanban was initially developed in the Toyota automotive industry in the 1970s and is today used as a control mechanism in modern plants worldwide. The basic concept of Kanban is the Kanban card where the card is attached to each part in the production flow. The Kanban card should only contain the information needed in the process and no more. By studying the movement of Kanban cards, the production rate and inventory levels can be determined. The number of cards is fixed and needs to be determined to find the optimal quantity in the production system. When a part is moved downstream or is delivered to a customer, the card is removed and sent upstream in order to authorise production of a new part (Xanthopoulos, Ioannidis & Koulouriotis, 2018).
Kanban board
Kanban is today used in many other industries than the automotive industry as it is well suited for processes that are repetitive. It is used in areas ranging from retail inventory to information technology sectors and especially for software development. Kanban is Japanese for signboard and a Kanban board is used to visualise and organise work items. Here WIP can be visualised and the progress of each task can be tracked (McLean & Canham, 2018). Emphasis when discussing Kanban is often put on Kanban as a part of Toyota’s Kanban system in manufacturing, little emphasis is put on the board itself which is often used in many areas other than in manufacturing contexts.
A Kanban system usually consists of a signboard where the cards are moved around and has over a longer time been used as an efficient way of visual management. The Kanban board is essential at daily
meetings; this gives a good overview and control over the workflow. It can be applied in many different work environments and has been used in agile software development. All the team members meet daily around the board and have a shorter meeting, usually around 15 minutes. At these meetings the team members discuss what was accomplished the previous day, today’s plan and if any problems have arisen. This kind of gathering contributes to teamwork and the ability to solve problems, which improves the productivity within the organisation.
9
A primary Kanban board commonly shows work that is accessible, work in progress and completed work. An example of this is shown in figure 3. Depending on the WIP flow, the board naturally constrains the quantity of assignments. By setting a limitation for the WIP it generates a pull system, which is created with the principle to only start a new work when the previous is done (Powell, 2018).
Figure 3: shows an example of how a simple Kanban board can look like (Powell, 2018).
Limiting ongoing work with Kanban
Some of the core uses with a Kanban system is that it can visualise workflow and manage flow as well as limiting WIP. Finding the right WIP limits in the Kanban system is done based on empirical experience as the approach for finding the correct WIP limit is trial, modification and observation. However, the number of persons working a process and the number of items they should process simultaneously can be used as a starting point in choosing WIP limits. These limits are not constant and will need to be readjusted under changed circumstances. Having WIP limits offers many advantages as the unfinished work can no longer be accumulated leading to shorter lead times. By not working on several work items at the same time, a better predictability of the process flow can be achieved.
The size of the input queue is based on the throughput of the team. This queue is the handover point for work that is submitted to the team. In this queue, work from the backlog can be placed by an authorized person on a daily or weekly basis. A small input queue leads to more flexibility as the queue empties itself faster and the right work can be prioritised from the backlog.
As the Kanban system visualises the workflow it is possible to make problems visible, since the cards will be blocked (marked by a sticker for example) if a problem occurs. It is important that these problems are visualised so that a team can dedicate their time to solving these problems before starting any new work. A direct effect of the WIP limit in this case can be that the problems need to be solved before it is possible for any additional work to be started if the WIP limit is already reached in the column. If the WIP limits are set too high, work is able to flow through despite there being blocked work. The problems will then be visualised, but not necessarily solved immediately. It is also possible to discover bottlenecks in the process by using WIP limits and visualising the workflow. If there is a bottleneck, work cannot be collected from the upstream queue and work will accumulate (Leopold & Kaltenecker, 2015).
Metrics in Kanban
Metrics are essential to determine the performance of a system. With the use of metrics, it is possible to determine the current status and if implemented improvements are productive. These can aid in finding where and what the problems in the current system are and thus where improvements should be made.
10
Root cause analysis is then useful to find the causes of the discovered problems. The metrics should be chosen with care as it is not effective to measure everything just because it is possible. The metrics that should be used depend on the goals that are being pursued. A recommended start for chosen metrics when implementing Kanban are the ones that can determine the continuity of the flow. These are lead time, throughput and flow efficiency. These metrics can also provide information about the variability which can be caused by WIP limits set too high or by many blockers (Leopold & Kaltenecker, 2015).
One practical analytical tool for workflow and a fundamental part of Kanban is the cumulative flow diagram (CFD). With the use of CFD it is possible to determine the flow of the system. The x-axis
represents time, and the y-axis the number of work items added cumulative over time (Petersen & Wohlin, 2011). These work items are divided by process steps. Therefore, it is possible to determine WIP for every process step at a given time by reading the diagram vertically. Horizontally, the average cycle time can be decided. If a system flows well, the lines have a constant positive inclination and the width between the lines dividing the process steps are constant. If there are any deviations these indicate bottlenecks which for example can be caused by a change of WIP limits, blockers or changes in available resources. See figure 4, for an example of a CFD.
Figure 4: An example of a CFD that can be used for determining WIP and cycle times for the different process steps as well as determining how stable the flow of the system is.
Lead times should be monitored to determine if they have changed over time and become shorter. As the lead time is measured, the throughput can be measured as well. The throughput describes the number of work items that are finished in a period of time and should increase when implementing Kanban. To determine the flow efficiency of a system, the processing time, waiting time and blocked time for a work item is compared to its lead time in order to understand how much of the time is actually spent working on the item. The number of blocked cards and their blocked time should also be investigated, as this can lead to new ways for improving lead times and aid in problem-solving (Leopold & Kaltenecker, 2015).
Challenges with implementation of Kanban
Changes are necessary if things are going to be improved. Changes can however be received differently among those affected by them. One challenge when implementing a Kanban system can be that it is
11
difficult to change the culture and motivate staff to use new practices. Challenges may arise as Kanban is integrated with the already existing processes and the implementation may require training. Problems such as managing WIP limits and selecting the correct work items according to priority may also occur
(Ahmad, Markkula & Oivo, 2013). When selecting work items, these need to be chosen according to priority and not according to area of interest or expertise. If tasks are not chosen as they should, these will need to be delegated and the team will not be self-organised (Stray, Moe & Dingsøyr, 2011). Policies should be established to help guide what actions and choices should be made. It is important that these policies are made explicit and visible to the team (Leopold & Kaltenecker, 2015).
In order to deal with these challenges, it is important that people affected receive proper guidance and training. If the purpose of the implementation is understood it is easier to engage the workforce. The workforce can be further engaged by sharing gains and communicating wins. It is important that attention is given to what is working well. As change is a continuous process it is important to sustain the change and build on it by continuous improvements (Almanei, Salonitis & Tsinopoulos, 2018). In the beginning of an implementation it cannot be expected that everything works smoothly, and a lot of changes can be requested. Therefore, it can be rewarding to have a short feedback loop in the beginning as some requested and needed changes can be simple to fix (Björkholm, 2015).
2.8 Visual management
The main foundations in Lean management are visual management, teamwork and continuous
improvement. In manufacturing it is common to use visual management at the daily meetings, particularly at companies that are aspiring towards Lean manufacturing. By using boards and information systems as a way of communicating, the productivity in the company will grow. The boards used at the daily meetings are usually standardised and the information can be written manually or with technical means. However, by using a standardised board, the entire organisation may not be involved. Therefore, it can be beneficial to let the users of the board be included in the developing process, which is an essential part in the Lean manufacturing process Kaizen where the focus is on operational improvement. By implementing Kaizen it helps the organisation step by step to improve by trial and error. When constantly re-designing and improving the visual tools, it gives the operators a deeper and broader knowledge.
Manufacturing companies in Sweden that are applying Lean often use boards in the meetings where important factors within the organisation's values and strategies are visualised. The different factors are often divided into categories such as safety, quality and the delivery performance. In order to be able to accomplish the daily work, the operators need information regarding material and product flow,
machinery such as stops and maintenance. Information regarding quality issues such as deviations in the material or in the equipment along with information if employees are absent or require training is also needed (Kurdve, et.al., 2019).
Visual management has the functionalities to transform information in order to manage production systems and works as an interactive platform where workers can obtain and exchange visualised information. At first, topics such as operation, time, value and quality were a central part of typical management problems. Different management techniques as quality control and industrial engineering were established in order to solve these issues. Today other matters have a strong influence; topics such as sustainability and better work environment.
There are different sets of digital technology that can be used for performing visual management. In today’s industry, digital concepts, platforms and tools are used in order to increase the interactivity with
12
workers through visual management. The intention of applying visual management can be to detect abnormal activities, which can be achieved by implementing different tools. Digital visual management often helps to visualise future problems. New tools such as sensors and virtual glasses can collect data that are more representative towards the real world.
Visual management does not only involve technical means, it also involves how people obtain the visual information. This includes how we see things by looking and observing, how we hear things such as speech and reports at meetings. By the same principles it can be important in safety and health matters if an unusual scent is detected, while touching and holding objects can tell if the quality of operations is good. These are all senses that affect how information is received (Murata, 2019).
13
3. Current situation at QC AstraZeneca Sweden Operations
This section describes the current ways of working at the two teams (MU1 and MU2) where the field studies and observations have been carried out. To describe the current situation, field studies, observations and discussions have been conducted.
3.1 Lean at AstraZeneca
Efficient ways of working are required in today’s industry to meet increased demands by effectively utilising resources, and this is no exception for AstraZeneca where Lean ways of operating are applied. The AZ supply system (AZSS) is the common framework of AstraZeneca and a representation of how they are working with implementing Lean into the organisation. This production system can be visualised in the common Lean house, see figure 5. At the foundation of the house are the values of AstraZeneca. These values are:
● We follow science ● We put patients first ● We play to win ● We do the right thing ● We are entrepreneurial
The structure of the house is represented by Lean principles that AZ stands by. These principles include standardised ways of working, Just-In-Time and elimination of waste. Each and every one of these principles are important in order to achieve the desired purpose. Their priorities are SHE, Quality, Supply and Costs. SHE stands for safety, health and environment. These priorities help AZ in decision making and they are to be prioritised in the presented order. The roof is represented by the common purpose of AZ (AstraZeneca, 2020c).
Figure 5: The AstraZeneca Supply System illustrating the Lean ways of thinking and working for AstraZeneca (AstraZeneca, 2020c).
More efficient ways of working are also required within the QC organisation to be able to respond to changes. This means that Lean principles are essential to reduce WIP levels and achieve more predictable and shorter lead times.
14
3.2 The process for a sample
A sample is a representation of a produced batch, and a number of analyses need to be performed on the batch samples to ensure that the quality of the batch meets the requirements. The sample is first received in what is called the ‘sample receiving’. Here the sample and the associated information is being
registered in a quality management support system. The information is then entered into the non GMP regulated planning system E-plan by the sample receiver. The samples are then being put in the ‘sample queue’, where they stay until an analyst starts analysing the sample. Depending on which article it is, different analyses need to be performed in order for the batch to be released. The analyses are performed according to different control methods (KM). The control method describes how to perform the analysis. Analysts need to be trained to perform the different control methods and are only allowed to perform analyses according to a specific control method if they are trained and qualified in the method. The information that belongs to a sample is:
● Sample number: a number unique for the sample (e.g. 0000001)
● Batch number: a number or denotation that is unique for the batch (e.g. ABCD) ● Article number: a number or denotation that is unique for the article (e.g. 123456789) ● Notation: describes which article it is (e.g. Product X, Substance X)
● Due date: the date which the batch needs to be completed for release (e.g. 2020-03-31)
● Control Methods: describe which methods should be performed for an article (e.g. KM0000001, KM000002). They always start with KM (short for ‘kontrollmetod’, Swedish for control method) and is followed by a series of numbers that are used to identify the specific control method. Before the analysis can be performed there is often need for preparation work such as gathering materials and glassware. Some analyses also require that solutions are prepared. The standard solutions should always be available. For the standard solutions, a Kanban system is used. A green card means that it is in-stock and a red card means that it needs to be replenished. For every solution there is a set number of cards used. The solution preparation part of the laboratory is manned to make sure that the standard solutions are available. Some non-standard solutions need to be ordered. For these solutions, a visual planning system is in place. Here the analysts place an order of the solution by placing a card on a board. Those working at the solution preparation then knows that the solution is requested and for when it is requested, by the information provided on the card.
When an analysis has been performed according to the control method, the results of the analysis needs to be evaluated and the results reported in the quality management support system. The evaluation of the analysis is then submitted for review. The review takes place to ensure that the analysis has been conducted correctly, that the documentation is correct and that the results are within specified
requirements. The reviewer also needs to be qualified to review and cannot review an analysis conducted by themselves. If the results do not meet the specified requirements, a deviation report must be
documented and then internally investigated. The investigation may require that the sample is re-analysed. All the analyses that are required to be performed on a batch need to be evaluated and reviewed before the batch can be released by QA, see figure 6 for the QC process for the samples.
15
3.3 Current planning activities for samples in QC
E-plan is a non GMP electronic planning system inspired by the principles of Lean. This system can be used for visual planning of the work and for supervising the current batch queues and resources. It is used by analysts, team leaders, managers and industrial engineers within QC. E-plan consists of several linked Excel files which can be used for different purposes. These files generate information from one another through linked cells. It is currently used for planning the analysing of samples in QC, to monitor the status of their operations and to obtain and generate statistics from their operations (AstraZeneca, 2018b).
The journey of a sample in E-plan
When a sample arrives to the QC lab it first enters the ‘sample receiving’ where it is being registered in the quality management support system. The information is then being manually entered into the planning system E-plan. In this step the sample is being registered in the sample receiving file by the sample receiver. Here the article number, batch number, sample number, due date and any other information such as if the batch need to be prioritised is entered, see figure 7 for a schematic example view of how the file is structured. Information such as notation and which team the article belongs to is retrieved from a linked file, the basic data file. From this file, information of which control methods should be performed on each article is retrieved. This file is like a database where information of all the articles is stored for the
laboratory’s teams, see figure 8 for a schematic example view of how the file is constructed. In order for a sample to be entered in the sample receiving, it needs to be registered in the basic data file.
Figure 7: An example of how the sample receiving file is structured, and the information that is being entered and retrieved from the basic data file.
Figure 8: An example of how the basic data file is structured. Here information of which control methods are performed for each article is found, as well as which team performs them on the article (color-coded).
16
When the article is registered, the registered batch and the associated control methods (KM) which are to be performed on the batch are displayed in another file, the analysis file. There is one analysis file for each team and here it is easy to see all the batches that are currently waiting to be finished. Here the status of all the batches can be monitored, to see which have been started and which still have control methods left to be performed. They are sorted according to due date and prioritisation, so the analyst knows which batches need to be prioritised first. If the batch is marked as ‘prio’ it will be shown at the top of the list. When an analyst retrieves the sample to start an analysis, the associated control method for that batch is marked as ongoing by the analyst. As soon as a control method is marked as ongoing for a batch, that batch is counted as in progress. When the analyst is finished performing the analysis, it must be reviewed, and it is therefore marked as being submitted for review. When a reviewer then takes it for review, it is marked as taken for review. When the reviewer is finished it is marked as reviewed. See figure 9 for an example schematic view of how the analysis file is structured for one team. Figure 10 shows the analysis file from E-plan for one team. When all the control methods that are to be performed on a batch have been done and are marked as reviewed, a date of completion for the entire batch is set. When a date of
completion is set, the batch is no longer shown in the analysis file as it is then finished, see figure 11 for the different statuses of an analysis. For some batches, the associated analyses are performed by different teams and the batch is therefore shown in several analysis files with different sample numbers.
Figure 9: An example of how the analysis file is structured, four batches are shown as well as the status of their associated control methods.
Figure 10: A view of how the analysis file is structured in E-plan (status for the different control methods as marked by analysts and reviewers: P=ongoing, G=submitted for review, T=taken for review and F=finished).
17
Figure 11: The different steps for the control methods associated to samples, from entering E-plan to being marked as finished.
The takt files read information about which analyses need to be performed from the analysis file. This is where information regarding the need of resources and turn-out is saved. This is currently used to plan the available resources. The analysts plan their activities themselves, by planning which control methods they are to perform and for how many hours. This is done in regard to how many batches are currently awaiting analysis with the assigned methods. The teams have daily planning meetings in the afternoon where they plan the following day and overlook the demand and resources.
There are also several files for tracking statistics in their operations. These files read information about the samples from the analysis file and the use of resources from the takt files. See figure 12, for a schematic view of the structure of E-plan.
Figure 12: A schematic view of the structure of E-plan, showing how the files are linked to one another.
Limitations of E-plan
The use of E-plan introduces a number of limitations in their operations, such as:
● Old data need to be cleared when new data is continuously being added due to constraints in capacity.
● There are more than 100 daily users and several users cannot edit the files simultaneously. ● Due to complexity in the linked files and the combination of large amounts of information, the
system is large and slow.
● Errors easily occur by users editing using short commands (copy and paste) when all formulas in the cell are adjusted. All users do not have the same level of knowledge in Excel.
● There is no constraint regarding how many batches that can be ongoing at the same time. Analysts can start analyses of several batches independently from one another, resulting in high levels of WIP.
18
The current planning system leads to several batches being ongoing simultaneously when analysts are allowed to start batches independently from one another, in regard to which methodologies they are to perform. This is illustrated in the schematic example of how the analysis file can be structured (figure 9). Here it shows that all batches are in progress and their respective analyses have statuses ranging from left to be performed to reviewed.
19
4. Configuration of an electronic Kanban planning system
The following section provides a brief description of what will be new practice for planning the analysis activities. The tested software is also briefly described, and it is presented how this software can be used to construct a Kanban board based on the existing requirements.
4.1 Introducing a pull system
To reduce WIP levels and improve the batch release rate, a pull system is to be introduced. All the analyses belonging to a batch should ideally be started at the same time. Batches with already ongoing analyses are therefore to be prioritised before any new batches are started. The analysts will use a decision tree as an aid to decide which activities should be performed, see figure 13. This decision tree has been developed to ensure that the correct activities can be prioritised and performed by the analysts, thus making the process policies explicit by specifying how the work should flow.
Figure 13: The decision tree to be used as a tool for prioritising work tasks.
To introduce a pull system, an electronic Kanban system is to be implemented. This electronic Kanban system will be used by the analysts instead of E-plan. As suggested, a Kanban board where ongoing work can be limited is constructed using the software Kanbanize.
Using a Kanban board along with the decision tree will then be the new practice of working instead of analysts planning their working hours according to which methodologies they are to perform and starting batches according to that. The team’s planning meeting is going to be moved to the morning, instead of having it in the afternoon the day before. Instead it will be planned which batches that should be completed and based on this it is determined which analyses need to be performed.
20
The priority will be to finish the ongoing analyses first, and thereafter take the ones that are submitted for review and if there are none, check to see if there are any ongoing batches with analyses left. The use of the decision tree will limit the WIP levels, since a new batch is not to be started unless there are no analyses left to perform on the already ongoing ones. This means that it will no longer be possible to start batches independently from one another, as the number of batches to be started daily will be limited.
4.2 Description of the software Kanbanize
Kanbanize is a Lean software program used for managing and visualising the workflow. All the activities in the work process can be divided in order to give a better overview of what is to be done and what work is under process. The two main components are the Kanban card and the Kanban board. Visualisation can be achieved by setting up the different work tasks on a digital Kanban board. This makes it easy for everyone working on the project to be continually updated. Every company can create their own individual boards to be suited for their business. Different kinds of cards can be created and linked together on the board in different sets of rules. The cards can be picked out by task, priority or colour. Kanbanize can keep track of deadlines and inform the users when a deadline is coming up by email communication. Kanbanize also has analytic tools where charts and diagrams can be made in order to analyse the workflow. The software can be integrated with other software programs in order to access other tools and services. It is possible to assign different permissions to the users (Kanbanize, 2020b). The pricing varies depending on the number of users and how many business rules that are applied
(Kanbanize, 2020c).
The Kanban board
The Kanban board is used to visualise the workflow and to monitor the status of all tasks. The default board has three different main columns visible: requested, in progress and done. These columns can be customised after desire to fit any specific needs. These columns are used to visualise where in the workflow the cards are located. Cards that are waiting to be processed are placed in the ‘requested’ column. These are usually organised in the order in which they are to be processed. When someone starts working on a card it is moved to the ‘in progress’ column. The card stays there until it is finished and moved to the ‘done’ column. WIP limits can be decided for columns separately or one WIP limit can be set for several columns by customising the board according to the needs. These limits are helpful when introducing a pull system as they ensure that the capacity is optimised and the workflow smoothened, by avoiding overloading. WIP limits are an essential part of Kanban and can be used to expose bottlenecks (Kanbanize, 2020d).
The board can consist of several workflows and these can be customised after desire. Every board has two default workflows: the Initiatives workflow and the Cards workflow, see figure 14 for a view of the default board.
21
Figure 14: Shows the default Kanban board and its workflows and columns
The Initiatives workflow is the top swim lane of the board where bigger tasks can be created and their progress tracked completely automatically as they go from ‘requested’, to ‘in progress’ and lastly to ‘done’. The Initiatives are bigger tasks or projects which can be broken down into smaller tasks (cards). These smaller work items get created in the Cards workflow. These cards are linked to their respective Initiative as child cards, and the Initiative is called the parent card. This parent/child relationship is useful to keep track of how a bigger project is going and the status of every work item in the project.
The Cards workflow is the second swim lane on the board and this is where the tasks and the status of all tasks are visualised as positioned Kanban cards. These cards can be linked to an Initiative in the Initiatives workflow, but they can also be independent (Kanbanize, 2020d).
There are two types of archives available: temporary archive and permanent archive. These are used to keep the board clean as well as to monitor historical data. To keep the board clean, all of the cards need to be moved to the ‘ready to archive’ column. They can be moved here manually or automatically by setting up time-based rules. This column is a temporary archive where the cards are kept for a certain time before being moved to the permanent archive. This time-period can be set to 1, 7, 14 or 30 days. The cards in the permanent archive are zipped and can be restored if needed (Kanbanize, 2020e).
The boards are collected in the team’s workspace on the dashboard. A management board can be created to make the status of several boards visible in the same place by showing the Initiatives workflows of the related boards (Kanbanize, 2020b).
The Kanban card
The card on a Kanban board is the representation of a work item. It carries the information necessary to identify the task. It can carry information such as title, Card ID, deadline, assignee (who is set to perform the task), size (how big the task is) and description of the task. The information that is relevant is entered when the card is created.
The cards can be colour-coded, have a type set that describes what kind of work item it is and have a priority. Tags can also be used to make it easier to search for specific cards using filters in the board view. It is possible to make card templates, where a card is designed to have a predefined set of properties. To
22
have properties beyond those already existing, custom fields can be created and added to the card to better present the information necessary.
Every card is assigned with a unique Card ID when it is created. This identification is used to refer to a card and can be seen as a serial number. This identification can be replaced with a custom one to match any other internally used numeral system (Kanbanize, 2020f).
There are also other types of card links than the previously mentioned parent/child link that can be used to represent other types of card dependencies. If several cards can be grouped together as the same type of work items with the same priority, these can be grouped together using a relative link. If there is a dependency with an order in which these work items need to be executed, these can be linked as successors/predecessors. The successor cannot precede the predecessor, and the successor will become blocked until the predecessor has been processed. It is also possible to synchronise properties between linked cards (Kanbanize, 2020g).
Creating Kanban cards
There are several ways of creating cards in Kanbanize. Cards can be created manually on the board, by filling in the relevant information for your work item directly and by manually creating links to other cards or Initiatives if desired. They can also be created over email by setting up email integration. Another way to create cards is by using the Kanbanize restful API (Application programming interface).
Cards can also be created in bulk by importing them from an Excel file. In order to successfully import cards, the spreadsheet needs to be formatted properly. The column headers in Excel need to be named according to the properties of a Kanbanize card such as title, deadline, custom card ID, type, size, priority etc. Custom fields can also be imported, and the property ‘Title’ is mandatory for a successful import. The column headers in the sheet are selected according to the specific needs. Cards can also be linked as a child to an already existing item (parent) by specifying its card ID under the column header ‘Parent’ (Kanbanize, 2020h).
Business rules
To automate processes and avoid as much of manual intervention as possible, business rules can be utilised. These are rules that trigger actions when a set of predefined criteria have been fulfilled. They are configured as “when condition X and Y is met, do Z”. The actions being triggered can for example be to create cards, update cards and move cards.
There are currently 16 types of business rules available that can be specified to describe several scenarios. These different rules trigger a predefined set of actions when the requirement of the rule is met.
The existing rules that can be customised to handle particular scenarios:
● Card is created: Triggers a predefined set of actions when a card is created
● Recurring create cards: Can be used to create new cards according to a specified schedule ● Card is updated: Triggers a predefined set of actions when a card is updated
● Recurring update cards: Can be used to update details of cards, e.g. change color of the card, according to a specified schedule.
● Card updated by email: Triggers a predefined action if a comment is added using the email integration functionality.
23
● All children are unblocked: Will unblock a parent card if all its child cards are unblocked ● Card is moved: Triggers a predefined set of actions if a card is moved to a specified
position/positions
● Child card is moved: Will move parent card to a specified position if any of its child cards is moved to a specified position.
● All children are moved: Will move parent card to a specified position if all of its child cards are moved to a specified position.
● Relative card is moved: Will move a card to a specified position if its relative card is moved to a specified position.
● Time-based rule: Will update/move/send notifications when a time-related rule is met, e.g. when the deadline is in 1 day, update the cards position on the board.
● Card revisions: When a card property is changed, this rule will save the card revision. ● WIP limit is reached: Triggers a predefined set of actions when the WIP limit in a specified
column/cell is reached.
● WIP limit is exceeded: Triggers a predefined set of actions when the WIP limit in a specified column/cell is exceeded.
● Card count: Triggers a predefined set of action when the number of cards in a specified column/cell is met, according to a chosen limitation.
When: This is the event that will trigger an action, e.g. trigger an action WHEN a card is created.
And: Here filters are defined to specify under which conditions the action is to be triggered, e.g. a card is
created AND it matches this position.
Then: Here the actions to be triggered, if the previous set of conditions are met, are selected, e.g. a card is
created AND it matches this position, THEN update the color of the card.
The rule created can then be given a custom name and a description for the scenario that the rule is applied for (Kanbanize, 2020i).
Analytics
Using the workflow analytics tool, different metrics can be tracked. These can be divided into five categories: Cycle time, Throughput, WIP, Flow and Forecasting.
● Cycle time can be tracked using the cycle time scatter plot, here it can be shown how long it took for a card to make it to the ‘done’ column.
● WIP can be tracked using the aging work in progress chart. By using this chart, it is possible to visualise how the cards are progressing towards the ‘done’ section of the board. Using the WIP run chart it is possible to monitor the daily WIP levels.
● To monitor the throughput, the histogram can be used to visualise the team’s productivity in the past. The run chart can be used to monitor the daily, weekly or monthly throughput.
● The cumulative flow chart can be used to show how many cards there are in respective column and the throughput of cards, sorted by date. This can be helpful in determining whether the process is stable or not over time.
● The heatmap analytics module can be used to decide where in the process changes should be made (Kanbanize, 2020j).
The advance search tool can also be used to find specific data. Here users can filter by particular search categories to generate the desired output and the filters can also be saved for future usage. The results can
24
be shown in different configured views and it is possible to generate reports. The filter can also be pinned as a widget to the dashboard, which is the entry point of the account, to be able to have quick access to important information. Examples of this could be to track Initiatives that have passed deadline on a particular board or to keep track of cards of a particular type. There are also predefined widgets that can be used to, for example, monitor the number of work items and their status (Kanbanize, 2020k).
4.3 Constructing a Kanban board at QC AstraZeneca Sweden Operations
The implementation of an electronic Kanban system poses demands, as there are a large number of different articles being controlled by several different QC laboratories. The system therefore needs to be able to handle this large number of articles and their associated control methods. It also needs to be electronically visualisable and easily accessible from anywhere, due to the extent and spread of QCs operations.There is also a requirement of minimal manual intervention from the analysts and team leaders. The transformation to an electronic Kanban system from E-plan should not result in a significantly more time-consuming and complex handling or planning of the samples logged in the system. The cards for the different control methods should not be created and linked manually as the effort for this would be too time-consuming. A process where all the needed information is entered manually and all links between cards are created manually would also result in a higher risk of handling errors compared to the planning system used today. Instead, the desire is to import cards using a system similar to the one being used today. By using the article number, cards for the control methods associated to the article should be created automatically. Information for that article should also be retrieved automatically when the article is imported from Excel. The only information that should be entered manually in Excel except the article number before importing, is the information unique for the batch and that is batch number, sample
number, due date and any potential remarks (such as if the status is set to priority). This is information that is being entered today using E-plan.
To start off, a configuration of a Kanban board with Kanban cards and business rules is set up for one team. This is done to present it to team leaders and analysts to gain their input. The aim is to discover potential hinders and problems that need addressing before an actual implementation is feasible
throughout the organisation. However, it also needs to be considered that the ultimate goal is to implement this at all of the QC laboratories, so it should be possible to scale up. When constructing and testing the board, consideration should therefore also be taken into how it could function with several boards and a larger number of articles.
The Kanban board
The Kanban board consists of two workflows: the batch flow and the analysis flow. The batch will function as an Initiative and it will be linked to all the control methods that need to be performed on the batch. The control methods will be cards in the analysis flow (analysis cards). The batches and its associated analyses will have a parent/child link.
The batch flow will consist of four columns: ‘batch queue’, ‘ready to start’, ‘in progress’ and ‘done’. The batches are created and imported into the ‘batch queue’ column. When a batch is ready to start, it is moved into the ‘ready to start’ column.
The analysis flow will consist of six columns: ‘analysis queue’, ‘ready to start’, ‘analysis ongoing’, ‘ready for review’, ‘taken for review’ and ‘done’. These columns have been created to match the current QC
25
process in order to visualise the workflow. The columns ‘analysis ongoing’, ‘ready for review’ and ‘taken for review’ are collected under one common section called ‘in progress’ since the status of the analysis is ‘in progress’ if placed in any of these columns. This makes it possible to set a common WIP limit for the ‘in progress’ section on the board. When a batch is created in the ‘batch queue’, cards for its associated control methods are automatically created in the ‘analysis queue’ by using business rules. When a batch is moved into the ‘ready to start’ column its associated control methods will automatically move to the ‘ready to start’ column in the analysis flow, see figure 15 for a view of the constructed board and its columns.
Figure 15: Shows the created Kanban board with the batch workflow and analysis workflow and their respective columns. Cards representing batches and analyses are visible in the respective flows.
The analyst starts a batch by moving one of the analysis cards into the ‘analysis ongoing’ column. When one of the analyses that is going to be performed on a batch is started and moved into the ‘analysis ongoing’ column, the batch card will automatically be marked as in progress as well and be moved to the ‘in progress’ column in the batch flow. When all the control methods to be performed on a batch are done, the batch will automatically move to the ‘done’ section, see appendix A for a more detailed illustration of the functionality of the board.
The idea is that only the batches that should be performed that day is moved into the ‘ready to start’ column. No batch should be in that column unless it is actually ready to start. The correct analyses will then be pulled from the ‘ready to start’ column and the ‘ready for review’ column in the analysis workflow from the analysts, as suggested by the decision tree. These two lanes will thereby act as the lanes from which work will be pulled. Which batches should be started is determined by an authorised person by moving them to the ‘ready to start’ column in the batch workflow. This should be done on a daily basis to avoid too many cards in the ‘ready to start’ columns in both of the workflows. When starting work, analysts should then primarily be working in the analysis workflow where the analyses to perform will be displayed. It is possible to hide and display lanes, columns and workflows to make it easier to orientate between the cards relevant at the moment.
The Kanban card
The board will contain Initiatives (batch cards) and cards for the associated control methods (analysis cards). The Initiative will show information such as batch number, sample number, article number, due date, notation and which control methods that should be performed on the batch. Three custom fields are created and named: article number, sample number and remarks. The field for remarks is left empty and
