Though Lean may seem as a toolbox of individual, independent tools, a holistic view quickly proves that all elements of Lean are intertwined. To achieve end-to-end flow, all elements must be balanced out. Besides throughput improvement measures, other factors such as workplace organization methods, total productive maintenance for increased resource availability, visual management aids for better communication and information flow must be considered. Together these are the building blocks essential for effectively tackling the Lean fundamentals: waste (Muda), overburden (Muri) and unevenness – variation (Mura).
In this post the focus lies on planning and scheduling, both activities at the heart of any enterprise. The first part deals with the distinction between planning and scheduling and application thereof in manufacturing organizations. In the second part I will look at the IT systems which are at the core of planning and scheduling tasks, including ERP Enterprise Resource Planning systems. Finally, alternative schedule execution systems and allowing for “pull-execution” of schedules of work, will be analyzed and compared.
Planning & scheduling
Planning and scheduling are decision-making processes used on a regular basis in many manufacturing and service industries. “Planning is a set of longer term actions that define the desired demand profile, capacity, inventory or response time required to deliver it, such that the purpose of the organization is satisfied” (Darlington 2018).
Planning and scheduling play an important role especially in procurement and production, distribution, information processing and communication of every enterprise. Planning and scheduling functions typically rely on techniques and heuristic methods to allocate limited resources to the demand. Pinedo avers that this allocation of resources must be done in such a way that the company optimizes its efforts and achieves its goals (Pinedo 2005).
In practical terms in the manufacturing industry, planning is about breaking down manufacturing orders into individual operation steps and defining a logical sequence of events that will deliver a completed product. A key element of this is understanding and managing the dependencies between events and resources. Literally speaking, planning is effectively defining “what is to be done” and “how”.
Scheduling is the “when” in a task. Scheduling factors in times and due dates to apply a chronological order to the plan, so it can be visualized in some form of timeline. “Scheduling is the production of a series sequenced and time sensitive instructions that will enable a work center, individual, team or any set of resources to perform within the overall plan” (Darlington 2018). It also means that operational steps are assigned to a resource, ensuring that it will be available at the required date and time. Scheduling effectively provides specificity and targets to a plan.
It is crucial to note, that the success of manufacturing planning activities is defined by the quality of data and formulated planning objectives. Manufacturing planning has evolved with the appearance of lean manufacturing concepts. As a result, strict adherence and enforcement of execution control of traditional planning and scheduling is creating conflicts in context with Lean. Replacing the traditional material staging principles of MRP e.g. “pushing” material to manufacturing stations by due date, Lean replenishment replaces this principle with self-regulating “pull” feedback control systems e.g. cards, bins. However, the importance of planning is not to be misconceived, even in an organization adopting Lean. Definition of planning objectives, measurement and decision support are evolving spheres of activity for planning in a Lean context (Bicheno and Holweg 2016).
The term ERP was first used by Gartner Group consultancy as an abbreviation of Enterprise Resource Planning. In the beginning of 1990 the term ERP was increasingly common and described the integration of business applications beyond singled out domains such as material resource planning, accounting, quotation and order handling. The use of ERP systems grew rapidly in the 1990s. The year 2000 problem aided this expansion greatly. Many companies chose to replace their legacy systems before the turn of the century. An ERP system is typically characterized as follows:
- An integrated system with various business applications
- Real-time or near real-time operation
- A common database that supports all the business applications
- A consistent look-and-feel across the different modules
Although many ERP systems evolved out of manufacturing resource planning systems (MRP) or production planning resp. control systems (PPS resp. PCS), some prominent examples (e.g. SAP) have their roots in finance/accounting systems.
MRP systems also evolved over time, from limited material resource planning (MRP I) to more integrated manufacturing resource planning (MRP II), developed in the 1980s.
As the ERP adoption in business grew, the complexity of the solutions and linked to that the implementation and operating costs of ERP software increased. Not only are on-premises hardware and software expensive capital investments, enterprise ERP systems often require the additional costs of custom coding, consultants, and training.
Meanwhile, ERP technology embraced the Internet. New features such as embedded analytics, mobile accessibility was added. Current systems are designed for cloud-based Software-as-a-Service (SaaS) delivery models, this allows for drastically reduced operating expenditures.
Advantages of ERP systems
- The prime advantage of any ERP system is that it allows central data storage and maintenance; thus, it provides one integrated single database for operations, sales and accounting data. This in turn enables users to share information between all components of the organization.
- Another advantage is that each module of the ERP system enters the same real-time database. This contributes to the avoidance of duplicate records or playback operations, i.e., redundancy is avoided. This results in improved data quality.
- Accurate and timely access to reliable information is a further advantage. This allows for time savings in producing reports or analysis of large amounts of data. In addition, an ERP supports companies to keep track of inventories continuously.
- Due to the integrative nature of modern ERP systems, process flows (work flows) are at the core of such systems. This enforces desired levels of collaboration between cross-functional departments and allows visibility of process progress and status. Process progress transparency in turn is advantageous when serving customers, as it ensures better information accurateness and improved response times.
- Professionally managed central data storage in an ERP system also helps to improve data security and availability. It further facilitates the assurance of regulatory compliance and helps in turn to reduce time and cost of litigation.
Disadvantages of ERP systems
- When implementing an ERP system, the current business processes of an organization must be rethought. Synchronizing the business to ERP process workflows is the key, otherwise expensive adaptions of standard ERP processes are required. ERP solution providers call this approach “best-practice-solutions”. This approach is highly questionable for many organizations as their business environment requires constant adaption and innovation. Consequently, many ERP implementations are too rigid for a specific organization and may even hinder its progressive advancement. A key determinant is the understanding of a corporation’s value generation flow. Typically, this is determined by analyzing the general flow of parts and products from raw materials to finished products. This typology is commonly referred to as V, A, T or X corporation or plant. V stands for diverging information and material flows, A for converging flows, T or X for mixed modes of diverging/converging flows. Especially V corporations with diverging information and material flows and specialized manufacturing capabilities will not be satisfied with traditional ERP systems (Darlington 2018).
- ERP systems in general are focused on financial and particularly cost tracking. As a result, the per se powerful tools of gathering activity-based operations data and combining them with financial parameters is often misused in ERP system to create a variety of financial indicators. Typically, financial measures are lagging indicators, as they measure the results of actions taken in the past. Though financial indicators are necessary for any business, they are often misinterpreted and lead to unwanted, or plainly wrong interventions into the operational business. Traditional cost accounting produces results that are misleading, at worst they can be entirely contradictory to Lean principles. They motivate people to use non-lean procedures, ultimately costing systems are wasteful. Standard costs methods can harm companies with lean ambitions, because they are based on premises grounded in mass production methods (Maskell 2017). An example: A large quantity of inventory carried at a higher absorbed cost was suddenly purged from the balance sheet and replaced by a smaller quantity at a lower cost. This is causing a sudden and unanticipated financial loss, when the excess and overvalued inventory was expensed. This is an illustration how short-term market valuation considerations can inhibit sensible long-term business decisions (Bell 2006).
- Variation is ignored by MRP. Fluctuation exists on various levels of a supply chain. Besides the common variation of demand, process and lead time variation are key influencing elements of MRP processes. Neither process nor lead time variation is considered by MRP appropriately. Process variation of a resource is commonly considered as an average value. In the MRP process common deviations from this assumed average create disparity in throughput, impacting on lead-time. Similarly, the behavior of procurement lead-times, results in deviations from the assumed average. Repercussions are material shortages, affecting customer deliveries or in contrast, create excess inventory and decreasing inventory turns (Bicheno and Holweg 2016). Within the same context there is the issue of rejects, or respectively “scrap”. MRP takes rejects as well as scrap into account as “factor” e.g. percentage of an expected total, unfortunately ignoring inherent variation. Consequences are the same as described above; potential starvation of a downstream process or unplanned for inventory.
- In most manufacturing environment, fixed lead times and infinite capacities are accepted principal parameters of MRP. One is aware that finite scheduling is available only in advanced planning systems and has its own pitfalls. Moreover, blind reliance on fixed parameters and typical human safety-thinking lead to the typical “increasing lead-time” syndrome. In this context, material supply is not taking place within the set lead-time; consequently, set lead-times are then increased to create a buffer resulting in more inventory being released into production. This in turn results in average production lead-times being missed. Resulting from this are schedule changes and subsequently even more potential shortages. Similarly, ERP systems hold capacity information. In more advanced MRP II systems, lead time calculations became possible. These systems are trying to balance key scheduling parameters such as Move, Queue, Setup and Run. However, even these systems fail as the calculated capacity load is a function of demand and lead time. As recognized, lead time is a variable parameter, depending on (arrival resp. demand) variation versus utilization of available capacity. Thus, the MRP II calculated capacity load is an approximation at best, in practical terms it is not meaningful at all (Darlington 2017).
- In a typical MRP environment, material supply is “pushed” to the consuming instances once they become available. Due to the underlying instability of supply (lacking adherence to scheduled receipt dates), the MRP outcome needs to be regenerated on a frequent basis, to continuously reflect the renewed current net-availability situation of materials. This continuous re-planning is not adding any value, but instead keeping large parts of an organization busy to continuously update all dependent requirement objects, e.g. purchase orders, manufacturing orders, etc. Scheduling and due dates in the past are a widespread occurrence in MRP systems. Hence, continuous manual intervention is required to forward schedule all replenishment elements to reflect a realistic planning situation. A further inability of MRP is to enable proper sequencing of work order queues. Simple sequencing based on due dates is not reflecting real life operation step progress, work station utilization and available “slack” time versus operation step time allocation.
Summary ERP systems
Though ERP systems are indispensable tools for any medium to large size organization, I fail to understand their marketed or sometimes even perceived status as savior of corporations. ERPs if properly implemented, can deliver tremendous value to an organization, but only if best practices are not sacrificed for static processes. Notwithstanding, most often forgotten, are aspects of usability and focus on data quality and governance that are the key ingredients to any properly functioning ERP system.
Scheduling approaches in manufacturing
A sensible measure for effective execution of schedules of work are those that satisfy customer demand with a minimum of work in process (WIP). However, low WIP bears inherent risks, such as inability to meet promised order due-dates. But the advantages of low WIP outweigh its risk. Maintaining low WIP reduces not only the net-working capital, but it also diminishes the order change-over risks in cases where customer requirements change. It further helps detecting loss -making orders earlier and foremost it avoids congestion related problems on the shop floor.
Taking this into consideration, depicting the service level achieved vs. WIP levels, permits an appropriate measure for effectiveness of execution of schedules of work. Whilst the WIP performance of the schedule execution of work is an important criterion, there are, incidentally many more parameters that must be taken into consideration to reflect dynamically changing conditions. Such changing conditions might emerge in the form of variation of demand, in cycle times, or set-up times. A good method of work schedule execution is also capable of readjustment to environmental changes. The ability of a method of work scheduling to compensate for changed conditions will be described as robustness in this post. Besides the traditional approach of MRP – Manufacturing Resource Planning, this post investigates alternative “lean” scheduling methods such as DBR – Drum Buffer Rope, CONWIP – Continuous WIP and the well-known KANBAN pull principles.
MRP – Manufacturing Resource Planning
MRP is a push method of production schedule execution. Though it is not considered to be “lean”, it merits being mentioned as it is widely used. It releases materials according to a master production schedule. Required work order release dates, respectively required material availability dates are determined through back-propagation of planned lead times. The planned lead-time of a production/work order is calculated as the sum of the set-up and processing times plus the consideration of time buffers, while uncertainties in quantities require planning of safety stocks.
The building of batches is offered on a production and delivery level, to aid the MRP to deal with significant set-up times or consolidation requirements at time of delivery. The changeover from customer demand to production/work orders is done within standard MRP procedures, which include demand/requirement netting, lot-sizing, time-phasing and bill of material explosion. The MRP run has to be carried out either at fixed times or is triggered by events that trigger subsequent activities, e.g. on the shop floor or in procurement. Shop-floor execution is typically managed with the help of dispatching rules (Hopp and Spearman 2012).
Kanban was originally conceived by Taiichi Ohno of Toyota. In comparison to MRP, Kanban is a pure pull system. Material consumption in a downstream work-step authorizes the setting up of a new production order for replenishment. Each material is controlled by a predetermined level of stock in circulation and linked to a production batch size, typically represented by the size of the holding bin. Besides production, Kanban is also used for goods transfer matters (move/withdrawal Kanban).
A working Kanban system adheres to the following principles:
- Each process issues requests (Kanban cards) to its suppliers as it consumes its supplies
- Each process produces according to the quantity and sequence of incoming requests (Kanban cards)
- No items are made or transported without a request (Kanban cards)
- The request (Kanban card) associated with a material is always attached to it
- Upstream processes must ensure defective items are not sent out. This to safeguard that finished products will be defect-free.
- Limiting the number of pending requests (Kanban cards) makes the process more sensitive and reveals inefficiencies.
The principal systemic difference between a Kanban system and MRP is that while in an MRP system information is brought to the material entry points and to the production units by the MRP procedure, the information in a Kanban system propagates upstream from work center to
work center at any point in time (in sync with demand resp. consumption).
CONWIP – Constant Work in Process
The basic notion of CONWIP is to ensure a constant level of WIP throughout the whole production throughput process. CONWIP releases input material in sync as final goods are needed. As opposed to Kanban, it is not a pure pull system. Both aspects, push and pull are incorporated in CONWIP. Kanban’s demand-driven replenishment is extended with the push approach of MRP. In CONWIP only a single set of cards is used for the whole covered process, while in Kanban individual cards are utilized between each pair of workstations. New demand triggers the release of new work orders. Work orders are listed in a common overview report. An authorization card is assigned to each work order. Each card remains associated to a particular work order until all work steps have been completed. The work item, once released, is pushed through the manufacturing system until the final product is completed in production. Production orders within a CONWIP line are dispatched by applying the FIFO rule ideally.
At the end of production, the card associated with the now finished work item is released. This allows a new work order to enter the production cycle. Input materials are released only if the WIP lies under the WIP cap. By applying this rule, WIP is not only controlled for each production step but for the whole production system. WIP remains constant (thus the name of CONWIP) as the total number of cards within the production system is sustained. The finished goods inventory is included into the WIP unless there is a constraint within the production system that restricts the throughput constantly. Attention needs to be paid to avoid constraint starvation. In the event of bottleneck occurrence, CONWIP allows for reduction of the total number of cards. On the contrary, it also permits increasing the number of cards which is conducive to raise WIP and ensure an expansion in throughput (Gastermann and Stopper 2012).
A further parameter in a CONWIP set-up is a work-ahead window. This anticipation horizon avoids producing finished goods whose due dates are too far in the future. Besides, such provision minimizes the risk of scrapping finished goods that degrade in case of changing customer requirements or for any other reason (Jodlbauer and Huber 2008).
Basic CONWIP systems as presented by its inventors Hopp and Spearman, are limited to single-flow shop floors only. More complicated layouts call for fragmentation to introduce several CONWIP lines or the adoption of additional dispatching rules. Although there is some similarity to a Kanban, the main difference is that the CONWIP does not mind the different types of materials or workstations, while the Kanban sets controls for every material on every single workstation (Spearman et al. 1990) and (Hopp 2008).
DBR – Drum Buffer Rope
Drum Buffer Rope – DBR, is a scheduling process focused on improving flow by identifying and leveraging the production system’s constraint. DBR is the operational production planning and control approach within the Theory of Constraints (TOC) developed by Eliyahu M. Goldratt and Jeff Cox (Goldratt and Cox 1984). The basic idea is to subordinate the production schedule to the system’s capacity-constrained resource (CCR). The rope is the pull system that connects the constraint resource to the upstream workstations and the gateway; the rope in controlling the release of new work, buffers the drum.
Typical variation in demand (statistical fluctuation) and the required production sequence (routing or in other terms dependent events) is absorbed by the provisioning of dedicated buffers. These buffers protect the CCR and the finished goods inventory against starvation. The buffer size is an estimation of processing and transfer times plus a certain amount of safety time. The loop between material release point and CCR is covered by the CCR buffer, while the loop between CCR and finished goods dispatch is controlled through a shipping buffer. To further safeguard against unnoticed constraints downstream of the CCR, a mechanism called buffer management can be put into action to set rush orders for the short-fallen materials. The CCR is the only work center where batching is used, as it is needed to optimize the use of the constraint resource (limiting change overs etc.). To further enhance the DBR system a second rope is linked from the finished goods buffer / shipping buffer to the first process step (gateway). This shipping buffer rope connects the entire system to market demand and limits overproduction (Jodlbauer and Huber 2008).
Comparison of Lean scheduling approaches DBR, CONWIP and Kanban
Drum Buffer Rope
|Outline||Stage by stage cards; production and move types of cards||Bottleneck protected by buffer; pull from bottleneck to gateway||Multi-stage pulls, linking end to beginning|
|Analogy||Supermarket||Herbie on the hike; Bottleneck determines the pace||Swimming pool circular lanes – one in, one out.|
|Additions||Capacity Kanban; ebans, golf ball||RYG FIFO queue monitoring at bottleneck and gateway||Multiple loops, separated by buffers|
|Strengths||Problems are apparent very quickly||Focus on bottleneck throughput||Fixed lead time; shifting bottlenecks automatically indicated by buffers|
|Limitations||FIFO routing; for a line, variation destroys throughput, resp. results in buffers; number of cards often changes||FIFO routing; loss of control downstream of bottleneck; potentially expensive inventory at buffer||FIFO routing in a segment; when bottleneck shifts throughput may be lost.|
|Schedule levelling||Yes, in a cell or line||Maybe, only if finished goods rope is tied to gateway||Probably, requires CONWIP planning queue overview|
|One-piece flow/batch||One-piece or bin||One piece or batch||One piece or batch|
Applicability of different schedule of work execution methods
Kanban works well with repetitive low variation of materials, e.g. component supply in leveled demand environment. However, Kanban requires a fixed flow path between stations. Additionally, it does not work well in environments with significant variation in processing time or multiple flows, resulting from separate routings for different products.
DBR – Drum Buffer Rope
DBR is best applied in an environment with shared (process) resources. Such resources typically operate towards maximum capacity load and represent constraints. The utilization of such capacity-constrained resource (CCR resp. drum) is to be maximized. Also, DBR requires a fixed flow path between the stations preceding the drum to the gateway. This concept is suitable for a limited variation of materials only, otherwise the CCR-buffer may take up too much space or tie up excessive inventory.
As opposed to Kanban and DBR, CONWIP allows for high variation of end products. It works well in make-to-order or assemble-to-order environments for customized products. It follows a defined path (routing) and due to capping of the inflow (release of work orders), the timing of the outflow (lead time) is rather predictable. Although it is lesser known than Kanban, it has many advantages in job shop environments.
Production scheduling and control strategies are an important part of supply chain management. The successful implementation of these strategies defines the ability of an organization to which degree it can meet customer demand reliably, with the least amount of spent input.
Manufacturing companies should try to adopt the schedule of work execution methods that best fit their individual situation. Kanban, DBR, CONWIP and MRP can operate in coexistence. Besides the described parameters of these methods, it is crucial to understand product typologies in general.
Converging or diverging product typologies are determining factors for shared resources particularly. Shared resources serve more than one value stream and typically involve changeovers in between. Shared resources need to be observed with caution, as they need to be scheduled separately. Shared resources may also need buffers, either in “divergence after” or “convergence before” scenarios. DBR, CONWIP and Kanban are therefore to be chosen in context of these guiding conditions always.