A look into the real needs of surface mount manufacturing to discover the opportunities that exist to improve optimization and performance through better planning of surface mount technology (SMT) lines, plus all related assembly processes.
Optimal planning and sequencing of work-orders is critical to achieve high productivity and on-time delivery especially where demand changes frequently and with short notice.
SMT production represents a unique challenge to existing planning and schedule tools. Performance of the SMT placement machines, no matter how efficient and highly performing individual SMT processes are, is directly influenced by the combination of materials setup on the machines and the choice and sequence of work-orders. SMT performance overall is dependent on the size of work-orders and the mix of products. Most of printed circuit board (PCB) electronics manufacturing has been high volume, where a single product is made continuously with few model variations; but today, it’s trending to high mix in which a wider mix of different products is made, ideally concurrently, but practically, and distributed across a few production lines. The entire factory capacity planning model is completely different for high mix because every element in the factory has to be split to interleave production of smaller batch sizes across the different products. The cost is the significant time taken for the changeover between products.
Companies who have made the transition unaided from once enjoying stable consistent high volume production, toward the more volatile higher mix production have recorded reduced productivity as much as 50% or more. Whereas high-volume lines can be expected to make several products within each five minutes, a low-volume, high-mix line will make that just the same quantity of products over a whole day. The production cost expressed in terms of each PCB produced, including the fixed costs of production, SMT machine investment, direct and indirect labor costs, etc., in the low-mix situation are more than 250 times higher than for high volume. The goal for any operation faced with increasing mix should be to find a way to minimize this cost and to achieve comparable levels of productivity no matter what the product mix or volume may be. This is as an essential factor for survival for many companies today.
Common tools used to create optimized production plans today are inadequate for SMT. Site-level tools are used in many cases, such as those provided as part of the site-level enterprise resource planning (ERP) operation, traditional manufacturing execution systems (MES), or specific advanced planning and scheduling (APS) solutions. These tools are designed for general manufacturing, and they are unable to take account of the specific characteristics and requirements of SMT. At the other end of the spectrum are tools provided by SMT machine vendors. The machine vendors are obligated to provide software that can create optimized programs for their machines, which have evolved to include support for using common feeder setups between products. The scope of these functions is limited because SMT machine vendors cannot be expected to provide factory-level solutions across lines with competitors’ machines nor for processes upstream of downstream of SMT.
Technology changes
Although the inadequacy of existing planning tools for SMT has been an issue for some time, the importance of it has recently been increasing rapidly. The origin of this is a reflection of changing demand patterns in the market. It is not only smartphones that consist of sophisticated and fashionable technology, with a high degree of variations and short product cycles. Demand patterns across the board in electronics are being influenced by rapid changes in technology, with high risk of deprecation of current models and the demise of the distribution chain in favor of direct shipping from factories, especially as Internet shopping and direct B2B ordering trends increase.
The control of product supply and demand in the medium and long term is not the issue. This continues to be managed by product managers and marketing teams, sensing market conditions and competitor activity, controlling demand using pricing and advertising. The short-term variation to the factory, however, cannot be managed in this way. With the very short or even non-existent distribution chain, the factory sees the raw fluctuation in short-term demand. To meet the 3changing demand, the factory has to decide whether to build to stock, effectively emulating the distribution chain and taking on the associated warehousing costs and risks of depreciation, or, whether to build products more closely in line with what the customer wants, addressing the challenge for planning to maintain operational productivity. Seeing how this trend is developing, and how it is linked to overall business opportunity, this has become a compelling need for many of today’s factories, to find a better solution for SMT shop-floor planning.
Optimizing SMT pocesses
While generic planning and scheduling systems can plan most manufacturing operations effectively, SMT represents a more complex optimization methodology, stemming from different aspects of the machines themselves and how machines are used together in a line. Each SMT machine has a theoretical placement speed, usually expressed as placements per second. This figure is presented in the marketing materials to indicate the performance potential of the machine. However, this speed can only be achieved under certain conditions.
In reality, many factors reduce the effective speed of the machine. These include factors that are usually considered unavoidable, such as certain parts that may need a slower machine movement because of their size or mass distribution or the parts may need visual recognition before placement. These losses can be minimized through the choices of machines working together in the lines to use the most effective machine for the job.
Other significant factors that are considered more avoidable reduce the effective speed of the machine, such as excess machine placement head travel across the PCB to and from the many feeders of materials. If the distances exceed certain limits, then the machine will take longer than advertised per placement. Because hundreds of materials can be set up at the machine side-by-side during production, inevitably not all of them can be situated close to the PCB, and it will take additional time for the machine placement heads to reach the more distant ones.
The machine program optimizer will try to minimize these losses by extended head travel, by putting the more popular materials within short reach and those used perhaps just once on each PCB toward the end. Different machine placement and movement technologies have been developed to mitigate these kinds of losses. However, many different factors need to be considered when fully optimizing machine programs because of the significant variations between machines with different technologies. Over the years, effective software available from the machine vendors has been developed to product optimized machine programs, including the Mentor Graphics Valor multi-vendor Process Preparation software. SMT production generally consists of lines of two or more machines because two or more machines are often required to provide the necessary number of material feeder positions. Machines with differing characteristics will be chosen, each specializing on different sizes and types of SMT materials, which provides the opportunity for the materials needed on a product to be placed by a machine with the most optimal characteristics. These machines may be ones working on a compatible software platform, or they could be from different machine vendors.
A key consideration for overall line optimization is the splitting of materials between the different machines. A key part of the decision-making process is then the balancing of work between machines in the line, a critical part of optimization, because the line can only be as fast as the slowest machine. The splitting of materials between machines and line balancing, where all machines are on the same platform, can usually be done by the machine vendor’s software. Otherwise, this has to be done manually, requiring much iteration, each after the individual machine programs have been optimized, until a satisfactory balance is achieved.
Compromisng SMT efficiency
For high-volume manufacturing, this machine and line-level optimization may be enough, but in all other cases, another much more important factor needs to be considered. The time taken to change from one product to another on an SMT machine can contribute more productivity loss than any other cause, even if there is only one model change per day. The tear down of the hundreds of materials from the previous product, and the set-up of the hundreds of new ones for the next product, in each case performing verification to ensure no mistakes in the material positioning, can take many hours. By contrast, the optimizing a couple of seconds off the time for a machine program or line balance can be insignificant by comparison for a smaller quantity of PCBs when operating in a high-mix environment. Un-optimized change-overs can reduce productivity by half, or worse depending on the number of changeovers per line per day.
The way to reduce the material setup times between models is to create “common feeder setups”. Instead of having a dedicated material setup for each product, a single setup is made to accommodate a range of products. For sequentially running products sharing a common feeder setup, all or most of the materials and their feeders positioning will stay the same; so in theory, a significantly reduced or even eliminated change-over time can be achieved.
Different elements of the common feeder setup can be used as required. However, the use of common feeder setups has a consequence. The order of materials on each of the machines is now determined by the combination of different products. The position of the materials still remains a critical factor that affects the machine program efficiency. Disregarding each product’s material setup makes it inevitable that each SMT program within the group of products belonging to the common feeder setup is going to be significantly slower, and significantly less efficient. The degree to which this is an issue is strongly determined by the degree of material commonality between the products that were grouped. Some optimization engines at the SMT machine level support the grouping of materials across different products, but, require a prior decision to have been made about which products should grouped together. Without thorough analysis of the commonality of all potential sequences of products that need to be made, together with consideration of the rest of the optimization parameters, the result will often be some poor common material setups in terms of machine efficiency capability.
Here then, we have a chicken-and-egg situation. Optimization of SMT, other than high continuous volume production, depends on both the optimization of the machine, line balance and material setups, plus, the sequence and timing of production to ensure that the right products will be made to meet delivery demand. Current planning tools are limited to require one of these two optimizations to be done before the other. In one scenario, the decision of the work-order sequence to satisfy demand is done first, determining which set of products should be grouped in a common setup, and then the common feeder setups within the group are optimized for the machines. In the other scenario, the best combination of product grouping is made from a machine performance point of view, and then, the work-order sequence decided which best fits the customer delivery needs. Whichever way it is done, the first decision becomes an assumptive input into the second. Neither kinds of existing legacy tools today, though specialist in their own areas can, for SMT, provide the complete optimization package in one. It can be very frustrating knowing that can be a better way to make each SMT plan, but being unable to do it, other than perhaps continuous manual trial-and-error.
Enhancing ERP
ERP system providers have been approached by many customers to improve site planning to take consideration of the SMT and more rapid demand changes from customers. However, the fundamental logic of ERP dates back to a time when mass production was the normal situation and production plans spanning months were easily created and reliably executed. Because of the lead-times of materials ordering, this basic business process is unlikely to change for medium- and long-term planning.
To address short-term issues, however, ERP and traditional MES systems simply don’t have the access to specific production configuration capabilities versus product models and other operational knowledge needed about the SMT machines and related processes. Therefore, ERP needs to be provided with some enhancement in terms of short-term planning, specifically engineered to create a flexible SMT production environment. It’s not another Y2K-type change, where reinvention of the whole of the manufacturing business systems would be required.
Enhancing ERP with a sophisticated SMT planning engine is a very manageable improvement to make from the operational and IT perspective. The Valor Production Plan solution is an SMT-specific optimization engine that combines the optimization of work-order planning, while simultaneously preparing the best materials setups for machine optimization. This essentially resolves the chicken-and-egg issue, finding every possible opportunity for real productivity enhancement in even the most highly volatile production operations.
Linking architecture
The Production Plan software is central to the manufacturing operation. Its strength comes from connections with key data sources. The first of these is the process engineering information. It is important to understand the exact detail of products that are to be made. An existing source from where this information can be derived is the SMT machine programs and libraries. In cases where machine vendors provide a software platform with an interface, data can also be extracted directly from these. The Valor Process Preparation software can also of course provide complete product setup information. The key is to understand which materials will be used, the way of setup and usage of the materials on each of the machines, and the quantities of each of the materials used per PCB. 7
Another key data source is the visibility of feedback of the operation on the shop-floor, to understand what is really happening and what the status is. This can often be the simple entry of the currently running production condition at the time of optimization, but can extend to a real-time feedback of progress from the line, using the Valor direct machine interface connections provided with the asset utilization or material verification software, for example. The more accurate and detailed the production progress information is, the greater the opportunity for risk-free optimization of near-term processes. Information about raw materials is also key. Interfaces between Valor Production Plan and ERP can be used to transfer available material stocks; however, ERP has limited knowledge of the shop-floor materials, so relying on ERP alone for materials management is not ideal.
Linking the Production Plan to an MES system where materials are more positively tracked on the shop-floor yields much better results and reduces the risk of material starvation when executing optimized schedules. The Valor Lean Material Management software can take this one step further, by not only providing a very accurate knowledge of material availability across all warehouses and shop-floor locations, but also providing just-in-time materials logistics. The use of Lean materials logistics means that existing work-orders can be much more easily cut-short without the need to count, remove, and re-assign the physical materials “pushed” on to the shop-floor. The link between ERP and Valor Production Plan is important for the transfer of the overall master site-plan form ERP and then subsequent feedback of completions. The Valor Production Plan software takes the master site plan and performs a detailed breakdown, creating an operational plan on a machine and line basis. Work-orders at the machine level are then used to associate and manage assignments of operational data, such as machine programs, material setups, and consumption information to be used as part of the simulation and optimization process. Once optimized, the resultant work-order sequences are ready to directly execute on each of the lines, following the time-line provided.
Operational improvement
Leveraging the power of the production plan solution has a profound effect on the shop-floor planning operation. The typical scenario happening every month completely changes. Originally, the shop-floor planning would start with demand detail coming down from ERP about the quantity required for each of potentially hundreds of different products needed in the coming period. Without knowledge of the specific SMT processes, the planning engineers would then need to spend many days trying to create a plan that would achieve the required delivery goals. Having time to provide a reasonable degree of refinement that considered alternatives was very limited, often being hampered during this process by last minute change requests. Often, many operations would need to be re-planned over and over again.
Shop-floor planning responsibility is one of the most critical issues to enable the factory to achieve its peak productivity. The production planning team is definitely amongst the hardest working in the operation; even though, they typically can only achieve the minimum to keep the operation running. Once the production plan solution has been introduced, the situation is quite different. The approach is now for the planning team to first ensure that data from all sources is updated and available. The planning team sets and adjusts the planning policy as controls within the software, specifying the relative importance of key planning criteria such that the resultant plan exactly meets the manufacturing needs. This can be set just once, although tweaking these rules can be effective where genuine compromises need to be made in response to genuine SMT capacity issues.
The difference made by introducing new planning tools is that not only is the commonality of feeders created and optimized by the system, but also selection of the groups of products and sequential order of the work-orders are decided by the system concurrently, so that the final schedule is the most optimum given the customer requirements and not dependent on any assumptions that simply were there to make the planning process easier.
by Michael Ford
Valor, Mentor Graphics