Why Improve Manufacturing Performance?
Manufacturing executives face increasing competitive pressure to improve manufacturing performance and supply chain management. Outsourcing, off-shore competition, eroding profit margins, and increasingly demanding customers are challenging traditional levels of performance and forcing manufacturing plants to either improve or perish. As such, producing a quality product at low cost no longer guarantees success in today’s market. The ability to build exactly what the customer wants and deliver it quickly is the new definition of competitive strength, requiring manufacturers to reach best in class levels in the following four areas:
Throughput: defined as the amount shipped from the plant. A focus on throughput allows manufacturers to measure something more meaningful—actual saleable products shipped—versus machine efficiencies or utilisation.
Cycle time: defined as the time to convert raw materials into delivered product. Cycle time improvements are a basic tenet of Lean because they directly attack waste- waste in inventory, queue time, and other non-value added aspects of manufacturing.
Customer service: measured with metrics such as On-Time-Delivery or Fill Rates. Metrics focused on customer service ensure that the most important characteristic of your operation- satisfying the customer – is achieved.
Inventory levels: including raw materials, work-in-progress (WIP) and finished goods. Inventory is a necessary evil because without raw materials or work-in-process inventory, manufacturers can’t make products. Still, idle and/or excess inventory is one of the most serious resource drains in a manufacturing operation, so inventory optimisation is a key element of lean as well.
Lean manufacturing concepts help companies improve cycle times and inventory positions, which in turn positively impacts throughput which increases customer satisfaction. Unfortunately, applying these concepts has proved to be difficult in many forms of manufacturing.
Complex environments: Lean’s Breaking Points
The bulk of successful lean implementations are in high volume, repetitive environments which is not surprising given its roots in automotive manufacturing. TPS techniques work well in high volume, low mix environments because of their inherent simplicity, but they have proven difficult to implement in less stable, more highly complex factories. Recently though, lean manufacturing has begun to expand beyond traditional high volume automotive environments into industries as diverse as pharmaceuticals, aircraft manufacturing, electronics, industrial products, and even office/administrative applications as practitioners are learning how to adapt the techniques.
Complex, asset-intensive environments differ from high volume environments in many ways as portrayed in the Table 1. Also, managing the flow of materials through a complex environment is different as well, requiring extensions to traditional lean techniques and terminology.
Challenges aside, due to the trend toward customisation and the proven benefits of lean manufacturing, it is worth the effort to modify standard techniques so that lean principles can be utilised in a complex environment.
Lean for complex manufacturing
Industry practitioners have identified seven areas where lean techniques can be extended to address the characteristics of complex manufacturing:
1. Flow Paths: Value streams for product families
Traditional high volume lean projects start by grouping products into a small set of product families, then creating value streams for each family. The approach involves designing cells that have equipment dedicated to the production of a single product family, and then implementing single-piece flow through that cell.
This approach is very effective in high volume environments like automotive where there are a relatively small number of products and dedicating equipment doesn’t pose a resource problem. Complex environments, however; have thousands of products visiting dozens of work centres using a variety of possible routings. As such, a more general approach to defining value streams is required to (a) accommodate a larger number of possible product families, and (b) enable equipment sharing among multiple product families.
Complex environments require a more general classification of product groupings that support the goal of establishing flow, but also allow equipment to be shared across product families.
We call this classification a ‘Flow Path’. Flow paths are a fundamental concept that allows complex environments to implement lean. By grouping products into families that visit similar pieces of equipment, flow paths provide a means to manage thousands of products through complex routings without requiring dedicated equipment. It facilitates the logical division of the plant into multiple flows, each of which can be considered a ‘focused factory’, independent of the others. Multiple flow paths can be defined for a plant, but a product can belong to only one flow path.
Flow Path Management (FPM) is defined as ‘the management techniques used to control the movement of materials through a plant’s flow paths.’ FPM concentrates on optimising the flow of materials through each path to maximise throughput and customer service while minimising inventory and cycle time. Flow Path Management builds on decades of research showing the benefits of focus factories, but extends this research to incorporate the lessons learned as lean manufacturing techniques have been applied in complex environments.
FPM shares several key principles with traditional high volume lean manufacturing and the Toyota Production System (TPS), including the elimination of waste and the application of pull scheduling. But while TPS was invented for high volume automotive production, FPM was invented for complex industries such as pharmaceuticals, metals, and electronics. One of the reasons for FPM’s success in these industries is that by breaking the process down into flow paths, some of the complexity of the problem is removed allowing visibility into the plant’s dynamics on a more manageable scale.
2. Organisation structures
Traditional high volume lean environments tend to organise production workers like they organise production equipment, dedicating people to a cell. They then balance work content among people in the cell to ensure capacity is matched to the takt time of the market.
But just as complex environments can’t dedicate equipment to a small set of product families, they can’t dedicate people, either. People must be more flexible, and able to work on a variety of products as demand shifts.
As opposed to the traditional structure of ‘departments’ and functional layouts, a complex plant seeking to embrace Lean needs to organise around the flow paths described earlier.
Flow Path leaders (similar to cell leaders in a less complex facility) can track the flow of materials through their flow path and (using the proper tools) identify and alleviate any impediments to effective product flow. With the organisation aligned according to these flow paths, metrics can be created and incentives administered enabling management to all levels of the organisation on a reasonable and actionable scale.
3. Performance Measures
Traditional lean relies wherever possible on visual, line of sight boards on the shop floor to track performance. These implementations, however has proven difficult to establish visual signals in complex environments; thus, performance measures in a highly variable environment need to track and communicate results of each flow path. Each flow path needs metrics that tell workers, collectively, how the entire flow path is performing.
Traditional metrics, like equipment utilisation or department budgets, are tuned to vertical silos rather than horizontal product flows. So companies implementing lean generally need to de-emphasise or eliminate traditional metrics and concentrate instead on the following three measures that track the movement of material to the customer through each flow path:
Cycle time: This is a measure of the time it takes for the product to flow through the factory floor.
Throughput: Throughput is a measure of the production rate (of saleable product) of the factory.
Delivery Performance: This is a measure of the ability of the factory to meet its delivery commitments to customers. It is a key indicator of customer satisfaction and has a direct impact on revenue generation potential.
There are two primary differences between these metrics and traditional ones. First, compared to traditional non-lean environments, these metrics emphasise product flow and customer value instead of efficiency and department-based metrics. Second, compared to traditional high volume lean environments, these metrics utilise software to collect data and communicate performance measures instead of relying solely on shop floor boards and visual signals.
4. Pull Scheduling
‘Pull Scheduling’ is a demand-driven method of material release. New material is produced only after existing material has been consumed. Pull scheduling contrasts with older methods that allowed computer programs to ‘push’ material onto the floor- regardless of whether there was room or a need.
Traditional high volume lean implements pull scheduling using Kanban cards. Each card includes a part number and quantity. When a product is sold or consumed, the card is moved to the prior operation, where an operator produces the specified quantity of that part and then moves the inventory with the card attached to the next operation.
For obvious reasons, this gets quite complicated in a complex environment. Inventory for each part in process is on the floor at all times, even if there is no immediate demand for that part and there are thousands of parts. Variability in demand dictates that appropriate levels of inventory for each part will vary over time, requiring management to add or remove cards quite often. Finally, work centres will frequently see different mixes of products. Bottlenecks will likely shift from one work centre to another leaving the number of cards assigned between pairs of machines out of balance, risking starvation of the new bottleneck.
Given the issues with traditional Kanban, practitioners have developed several alternative forms of pull scheduling better suited for complex environments. While the techniques are known by a variety of names such as ‘Generic Kanban’, CONWIP, Drum Buffer Rope, or POLCA, they all share common techniques to (a) have the ‘card’ represent a flow path or generic family of parts, (b) have the ‘card’ able to float more freely to buffer the bottleneck from starvation, and (c) reduce inventory while still protecting throughput from sources of variability like product demand or unplanned equipment downtime.
5. Bottlenecks and capacity planning
It is no secret that excess capacity (extra machines or staffing) can hide waste and mask many evils within a plant. These large safety nets can be extremely costly in terms of tying up cash, increasing cycle time, or requiring an unacceptable outlay of capital.
Traditional high volume lean uses rules of thumb and simple calculations to determine the best levels of capacity utilisation. But in complex environments, setting optimal capacity levels cannot be accomplished using simple rules of thumb. With hundreds of products, each with varying demand and routings, bottleneck resources shift. These bottlenecks constrain total throughput, and management must identify them as the first step in optimising performance, a technique popularised by ‘The Goal and the Theory of Constraints’.
Calculating capacity utilisation must also be sensitive to variability in equipment processing times, downtimes, setup times, and product demand. Variability, after all, is the one of the primary reasons inventory and capacity buffers are needed. By incorporating variability in its calculations, plants can set capacity levels high enough to provide a buffer. Moreover, they can calculate which sources of variability should be attacked first to reduce the need for extra capacity buffers.
6. Inventory Optimisation
Inventory is another buffer that can hide waste and mask variability within a plant. As with capacity utilisation, traditional high volume lean uses rules of thumb and simple calculations to determine the best levels of inventory. For example, inventory levels are set in most Kanban implementations by using calculations based on Little’s Law—inventory equals throughput times cycle time. Kanban card counts are initially calculated by multiplying a cell’s takt time by its expected cycle time. Over time, management slowly reduces the card count to lower the inventory level and expose and fix production problems hidden by that inventory.
In complex environments, setting optimal inventory levels cannot be accomplished using simple rules of thumb. As variability in demand, product mix, setup times, process times, and machine reliability increase, the duration that a set inventory level is valid becomes increasingly smaller, making a dynamically calculated value necessary. Advanced approaches provide updateable optimal levels, and the same concept holds true for determining optimal inventory policies for raw material or finished goods inventory buffers.
The following graph illustrates the point with a simple example. Cycle time for a product increases as capacity utilisation increases. The line marked ‘high variability’ shows a plant that has high variability in product demand, processing times, setup times, and equipment downtime. The line marked ‘low variability’ shows a plant with identical average product demand, processing times, setup times, and equipment availability, but with lower variability in these values. By using operations research to draw these two lines, one can see how the cycle time for the low variability plant is much shorter. Management can use a graph like this to decide how much capacity utilisation, cycle time, and inventory are optimal for their current levels of variability.
7. Lot Sizing
Lean thinking promotes single piece flow. But when setup times prevent lot sizes of one, parts must be produced in lots that share a common setup. (The term lot size is often called batch size or campaign sizes. The terms generally mean how much a product should be produced each time it is produced.)
Traditional high volume lean uses rules of thumb called ‘Every Part Every Interval’ (EPEI) to set lot sizes. Again, because of variability, this approach is unworkable in complex environment. It suffers from two weaknesses. First, the appropriate target capacity utilisation needs to be optimised based on variability. A plant with high variability might optimise utilisation at 70%, while one with low variability would perform better at 90%. Second, the correct lot size for each part will vary as product demand varies and as production capacity varies. New operations research methods have provided techniques to optimise these lots sizes, taking multiple sources of variability into account in their calculations.
For over 30 years, high-volume, repetitive manufacturing environments have benefited from lean manufacturing methodologies. Unfortunately, those techniques did not translate well to a complex environment with huge numbers of products and high variability in products and processes.
Lean manufacturing techniques can be adapted to complex environments when the effort is focused on:
Changing traditional metrics and measurements of performance
Using flow path management to derive more flexible approaches to defining value streams and organisational structure
Utilising alternate means of calculating inventory, capacity planning and lot sizing
While many lean manufacturing techniques can be manually implemented in high volume, repetitive environments, processes in complex environments can become much more complex. Manufacturers can benefit from using available software solutions for implementing flow path management in a complex environment to take advantage of lean manufacturing techniques. ☐