The Lean Advantage

Filippo Focacci and David Simchi-Levi discuss how the process and hybrid industries can benefit from the implementation of lean manufacturing principles.

Lean manufacturing is one of most important business processes in manufacturing today. Started in the mid 1940s as Toyota was trying to catch up with US car manufacturing companies, it was designed and implemented by visionaries during a period of about 30 years. This Toyota Production System (TPS) relies on two important concepts, Just-In- Time (JIT) and Autonomation, or “automation with human touch”, and its goal is to remove any form ofwaste from the production system.

In the mid 80s and early 90s, many automotive manufacturers followed Toyota’s leadership and tried to replicate the benefits of the Toyota Production System. In the last few years, other manufacturing industries, including electronics, office furniture, telecommunication equipment, have implemented lean very successfully.

Not surprisingly, most successful lean implementations are found in the discrete industries characterized by high volume, repetitive operations and assembly lines. As a result, lean is an excellent fit to some of the challenges that these companies face.

Unfortunately, there are fewer applications of lean manufacturing in the process and hybrid industries, in part because there is not a natural fit between lean and these industries, and in part because manufacturing processes in these industries tend to be very complex. Indeed, it is difficult to adapt lean manufacturing strategies that have been implemented in discretemanufacturing to the process and hybrid industries.

Challenges in process
The process and hybrid industries face some important challenges. These include:

• High demand variability – often generated by promotional activities, new product introductions, product phase-outs and interactions with large retailers.
• High pressure and strict constraints on product quality – generated by various regulatory compliances (e.g. sanitary regulations) as well as shelf life constraints.
• Strong focus on cost reduction and return on asset – manufacturing and supply chain costs are often the most important competitive dimension for companies producing commodity products.
• Higher frequency of schedule changes – today the expectation is that planners will change schedules on a daily basis, which generates an increase in setups, manufacturing costs and introduces production variability. • High pressure to increase service level while reducing inventories.

As illustrated in Table 1, these challenges are quite different than those faced by the automotive industry, or more generally by discretemanufacturing companies.

It is easy to see that implementing Toyota’s lean techniques not only can be very challenging in the process and hybrid industries, worse, it can generate very poor manufacturing performances, if not implemented correctly. For example, reducing lot sizes in the may lead to low utilization of tank capacity and hence is not effective.

More generally, it is not clear how to implement a pull strategy in a production process that possesses strong economies of scale and/or requires batching.If raw materials and intermediate products are stored and produced using tanks, it becomes very difficult to implementpull scheduling and a one-piece flow process. Similarly, the kanbanconcept is appropriate for discrete manufacturing, but impossible toimplement in the process industry.

Quick changeovers can be very ineffective, difficult or even impossible to achieve in the process industry. For example, in the food industry a flavor change may generate a “purge”, which generates a waste of product. In the pharmaceutical industry, a “cleaning-inplace” may require a full day. In the chemical industry, a changeover may require changes in the configuration of the plant.

Achieving smooth production may also be very difficult when some products require high volumes and other low volumes or when the demand is volatile. Machines cannot be dedicated to a single product family and the bottlenecks continuously float from one area of the plant to another. Computing capacity utilization and inventories with simple rules of thumb based on takt times or “zero inventories” are also not appropriate.

Thus, classical lean techniques cannot be applied to the process and hybrid industries. However, as described below, lean principles can be applied to identify requirements for implementing effective manufacturing strategies in the process and hybrid industry:

• Take a holistic view Sequencing and scheduling decisions must be made by looking at their impact on both resource efficiency and inventory levels and including information about raw materials, intermediates and finished goods.
• Attention to details Operational details must be taken into account when generating production plans and schedules. Building a plan that ignores key manufacturing constraints such as tank capacities, cleaning rules or sequence-dependent changeover times will generate continuous adjustments to the plan. Infeasible plans generate many unplanned changes, which create more changeovers and waste. Additional changeovers generate loss of capacity and inventory shortages, which in turn will generate expedite orders and new infeasible plans.
• Control WIP Controlling and reducing WIP can be achieved by carefully coordinating intermediate product production, finished good production and demand signal.
• Reduce cycle time Cycle time includes two components: wait time and processing time. Wait time is reduced by better coordinating the flow of material (e.g. intermediate products and finished goods), while processing time reduction is achieved thorough optimally balancing changeover times and costs and inventory costs.

These requirements show that both the old MRP (materials requirement planning) logic and some simple rules that work well indiscrete manufacturing become inadequate when trying to achieve operational excellence in the process and hybrid industries.

The requirements also suggest that there is a need for effective optimization methods that take a holistic view of the production process, balance the various trade-offs while considering all business constraints, and provide planners with enough flexibility to easily and effectively modify production plans.

Putting into practice
The case study that follows is based on a software implementation at a customer in the Fast Moving Consumer Goods (FMCG) sector. The study is based on the comparison of production plans manually generated by experienced planners against production plans generated by the same planners using “Plant PowerOps” on the same data.

ILOG Plant PowerOps (PPO) is an integrated planning and scheduling application designed to meet the complex challenges of the process and hybrid industry. Bridging the gap between ERP and MES systems, it aims to improve companies’ flexibility and real-time responsiveness.

The customer’s manufacturing process is a relative common process that can be found in the Food and Beverage, Pharmaceutical or Chemical industry: intermediate products are built in production tanks and continuous equipments and stored in storage tanks. Finished products are produced in filling and packaging lines. Chocolate, tobacco, ice-cream, drugs, detergents, are all examples of products that share similar manufacturing complexity:

• High product mix, with new product introductions and phase outs, very volatile demand and high service levels requirements
• Shared resources, both for production and for cleaning
• Tight shelf life and maturation time constraints both for intermediate products and finished products
• Multi-step recipes in continuous equipments, production tanks and storage tanks
• Cleaning-in-place and traceability regulation

The main objectives of the project were to reduce product waste and cycle time and to improve planner’s ability to make optimal decision faster than before. Clearly this customer cannot meet these goals by applying traditional lean techniques designed for assembly lines. Kanban cards cannot be used, setups are very time consuming, generate product waste and cannot be reduced. Finally batch sizesare tightly connected to the physical capacity of the tanks.

As explained earlier, the appropriate approach to achieve these goals is to look at the lean principles rather than looking at the lean techniques, i.e. translate the principles into a different set of techniques:

1. Decisions on production quantities, WIP and stock coverage should be tightly linked to sequencing and scheduling decision. Using a single integrated planning and scheduling model planners are better able to synchronize manufacturing execution with demand signals and to find the right trade-offs between operational efficiency, asset utilization, inventory coverage and service levels.

2. Since kanban cards cannot be used, the coordination of the material flow and minimization of WIP is achieved by planning and scheduling both intermediate products and finished goods in tight coordination.

3. In order to generate a schedule that is executable on the plant floor, the key manufacturing constraints need to be taken into account by the planning and scheduling system.

4. The planning and scheduling system must enable planners to quickly generate new plans, analyze the quality of a given plan, and to change it by easily overriding the system’s decisions.

After implementing Plant PowerOps, the results were analyzed in terms of inventory coverage, operational efficiency, andproduction smoothing.

Inventory & WIP
The customer uses inventory targets expressed in minimal and maximal days of supply. An excess inventory occurs every time the inventory position is above the maximal days of supply, while an inventory deficit occurs every time the inventory position is below the minimal days of supply.

The optimal inventory position in a constrained environment will tend to move from the minimal level to the maximal level as regularly as the capacity constraints and changeovers will allow to, with a saw-toothed pattern.

In order to evaluate the impact of the PPO implementation, for a given set of SKU, the inventory excess and deficit was measured and compared against the inventory excess and deficit of a manual plan.

The results (depicted) clearly demonstrate a strong reduction of inventory excesses (which went almost down to zero) and deficits. Plant PowerOps was able to achieve these results by finding the right balance between inventory costs and changeover costs under tight resource capacity constraints.

Note that the decrease of inventory excess not only leads to a reduction of inventory costs, but it also leads to a reduction of the risk of product waste as these products are subject to obsolescence.

The reduction of inventory deficit is also important. Using PPO, planners can now better respect the safety stock targets, and these targets can now be recalibrated to manage unexpected eventsduring execution

Operational efficiency
In terms of manufacturing performance, the most important result was a good improvement in operational efficiency. Roughly speaking, operational efficiency is defined as the ratio between the operational time and the net production time and it provides a good measure of how well plant equipment is used.

When a plant runs at very high capacity utilization, any increase of operational efficiency directly translates into an increase in throughput. More important, if a plant runs at very high capacity utilization, and the market is growing, any increase in operational efficiency generates an increase in sales and a delay in necessary capital investment.

As mentioned before, the reduction of changeover time is a key component for the reduction of cycle time. In fact, cycle time is made of two components: wait time and processing time. The wait time is reduced by better coordinating intermediate products and finished goods, while the processing time is reduced by reducing the changeover times.

The results obtained (see Table 2) illustrate a significant improvement over the manual process.

Production smoothing
The ability to achieve a smooth production even in the presence of high demand variability is critical in creating a lean production system. In the process and hybrid industries, production smoothing should be measured both in terms of production frequency and production volumes.

In the ideal situation, high volume products should be produced in fixed quantity and fixed production wheels. This is unfortunately not always possible because high demand variability and physical batching constraints of intermediate products make fixed production quantities infeasible, and high changeover costs pushes low and medium volume products to be interleaved with high volume products.

Depicted below is the production variability (in volume) of different SKUs. Production variability is defined as the ratio between the standard deviation and the mean of production quantities. It is easy to see that the optimal plan has a lower variability, thus leading to better utilization of resources. The production frequency of the optimized plan is also significantlymore stable than the frequency of the manual plan.

Closing the gap
Overcoming the problem of MRP, lean methodologies have helped discrete and repetitive manufacturing industries to better synchronize demand signal and execution, to produce high quality products at lower costs. Although traditional lean techniques are inadequate for the process and hybrid industry, lean principles can be applied here but require a different set of tools.

While the first generation APS systems still mimic the MRP logic and do not take into account plant floor constraints, manufacturers can now benefit from a new generation planning and scheduling software that were designed to help planners dealing with the complexity of high product mix, shared equipments, physical batching constraints and high regulatory compliance constraints.

Plant PowerOps helps implementing lean principles by providing an integrated planning and scheduling model that makes the right tradeoffs between supply chain goals and manufacturing goals.

Taking into account key manufacturing constraints, the gap between planning and execution can thus be closed – enabling planners to generate plans and schedules that are really executable in the plant floor.

Dr Filippo Focacci is Product Manager, Plant PowerOps, ILOG; David Simchi-Levi is Chief Science Officer, ILOG, and Professor ofEngineering Systems, MIT.

----------------------------------------------------------------------------------------------------------

----------------------------------------------------------------------------------

A Lean Primer

The main objective of Lean is to provide the best possible service to the customers through the elimination of all forms of waste. Waste can take the form of material or energy waste, inventory, defects or waste of capacity.

The elimination of waste requires a continuous improvement process in which workers are capable to quickly detect problems and are empowered to solve them effectively. It also requires a high stability of the production system achieved through production smoothing, that is, a manufacturing process focusing on a constant volume and product mix.

Figure 1 represents TPS’ principles, techniques and objectives. Specifically, Toyota had identified several techniques that helped the firm create one of the best run manufacturing processes in the world and achieve best quality and service atthe lowest cost.

For example, “kanban” cards are used to control work orders, to implement pull scheduling and to reduce work-inprocess inventory. A kanban card is associated with one piece of flow and is used to trigger new productions into the system. The number of cards controls the amount of WIP in a specific area.

A strong emphasis is put on production smoothing, through constant volume and constant product mix. This is achieved using “takt time”. Takt time is the frequency of consumer demand and the goal is to make sure this frequency is closed to the frequency of in which parts are produced in the system.

If the frequency of production is greater than the frequency of demand, then we start building inventory, i.e. waste. If the frequency of production is smaller than the frequency of demand, we have wait time, shortages and lowservice levels.

By monitoring takt times, Toyota was able to smooth production and define the optimal capacity utilization. In order to implement constant volume and constant product mix, a strong focus is dedicated to the ability to implement quick changeovers and to reduce lot sizes (ability to make every piece every day).

In this context, empowering employees and team work becomes necessary. In fact, the low levels of WIP inventories (buffers) uncover problems as soon as they arise. Workers must have the ability to stop the production line, and to start production again, problems need to be fixed as soon as possible. This not only leads to detecting problems very quickly, but it also fosters problem solving capabilities.

In fact, the ability to stop the production line and the need to solve problems very quickly is part of the characteristics of lean manufacturing. In this manufacturing strategy, flexibility becomes extremely important, and it is achieved through workers’ cross-training.

Guiding principles Lean techniques have been clearly driven by the requirements of assembly line manufacturing and tuned for the Toyota manufacturing environment. Achieving similar results in different manufacturing environments requires the understanding of lean’s four key principles:

• Take a holistic view
The production environment is a system that needs to be optimized as a whole. The objective is the close coordination of operations that are physically separated.

• Attention to details
Operational details are strategically important in lean manufacturing. The focus on setup reduction is a good example: instead of ignoring setups or taking them as a fixed constraint of the system, engineers try to reduce the setup time so that non-productive operations are minimized.

• Control WIP
Controlling WIP is an important objective that is achieved through the use of Kanban cards. The key idea is to change the manufacturing process from a “push based” manufacturing process to a “pull based” manufacturing process. Demand triggers production activities and gives the right frequency of production.

• Reduce cycle time
Reducing cycle time is a key objective in lean manufacturing and it is achieved by reducing setups, reducing delays, coordinating machine maintenance with production operations, and optimizing space in order to better utilize workers,equipments and workstation.

----------------------------------------------------------------------------------