Programming SMT lines by common materials usage can cause massive hidden waste.

“Lean,” as applied to production processes, often leads to change in the flow of an operation. The same can be seen when applying the principles of Lean thinking to manufacturing engineering, such as data preparation, SMT machine programming and production planning. Engineering processes are often refined over many years of “best practices.” They become critical processes, which, if not done correctly or completely, can threaten productivity and delivery, as well as the quality of finished products. These “silos” of engineering have become fixed. Let’s challenge this state, and determine with our application of Lean thinking what could be delivered by a different flow, free from the encumbrance of legacy practices. The result is a world away from what most people would expect, with significant potential benefit.

For an EMS company, the analysis of incoming product data is paramount to accurately understand the manufacturing requirements of new products and hence cost of manufacture. The data preparation process takes whatever data are provided from the customer to create a qualified product model. Unless the customer provides a complete and intelligent data set, much data reconstruction may be necessary and often several assumptions have to be made. For the OEM, we would imagine that because there is the opportunity of vertical integration between product design and manufacturing, the Lean approach in the use of an intelligent file is a no-brainer.

For all manufacturers, however, there is the issue of local bill of materials. The purchasing team at each local factory is on a mission to secure reliable material supply with the minimum of cost and lead-time, while maintaining quality. The materials chosen may not be exactly the same ones specified as part of the analysis of the PCB design layout. Running design for manufacturing (DfM) tools as part of the data preparation flow at the start of manufacturing, crucially using the same tools and rule-set as used during design, but with material information of what will be used locally, can provide another level of protection to avoid manufacturing issues, including defects, delays and unplanned countermeasures. Engineering then has a clear response to purchasing as to which materials can and cannot be used, without argument. For an OEM, this can make huge quality and productivity improvements, which also can be the case for an EMS, although discussions with the customer can be more complex, as the benefits of this feedback to them are shown.

The engineering data preparation team then sits justifiably firmly in control of incoming product data and purchasing material choices, both of which are qualified against production process capability, including the SMT machines. The final stage of this data preparation flow is usually to create actual SMT machine programs. Whether using multi-vendor third-party programming software tools or simply using the programming tools provided by the machine vendor, the final “feel good” factor comes when the result is a set of decent SMT programs for the product.

Planning restrictions. This, however, is where the problem starts. Several potential SMT line configurations could have been used for the product, but often because of time considerations only the most likely or intended configuration is selected and used for product qualification, which then ends up being in most cases the only configuration that product will be recommended or even mandated for. Often products with a similar PCB layout and materials usage will be grouped by the SMT programming engineering team, with all products in the group sharing a common material setup on the SMT machines, so that losses at changeover time caused by material changes will be reduced. Few people will question the wisdom of this flow because it is critical that the qualification of the product to the operation is complete before production can start.

This latter part of the process places serious restrictions on the choices offered to the production planning team, who somehow need to work out the sequence of products through the operation to meet delivery demands. A massive amount of hidden waste can be the result.

Consider a line that is either dedicated to a particular product or to a group of products sharing a common feeder setup. The SMT programming team will have optimized the SMT programs for the line as well as can be expected so it will run quickly and smoothly, for example, capable of completing 50 PCBs per hour. If the completion demand remains at 50 per hour, and ignoring operational stoppages, the line can be said to be running at 100% efficiency. If the delivery demand falls, however, the line will effectively overproduce.

A decision needs to be made as to how to handle this. Not running the line obviously has a significant cost of lost opportunity, but because of the dedication of the product group or single product to the line, any changes to the line also come with a significant cost. To avoid both scenarios, the decision is often to keep the line running, storing the additional products as semi-finished goods, while hoping the demand from the customer or at least the next production process will change, and exceed the 50 units per hour using up the excess semi-finished goods. Over time, if the average delivery demand does average out at 50 units per hour, then everyone in manufacturing will be happy, as the cost of storage, additional logistics and management, risk of depreciation, and risk to the products themselves is outside of their responsibility.

This often hidden or ignored cost is actually the best possible scenario. In real life, the average demand rate will not be exactly the same as the optimum line completion rate. Planning will typically be biased in favor of ensuring peak demands can be met, so often the problem is eventually the production line has to stop; otherwise too much excess stock will accumulate. The significant cost of storage and warehousing only serves to delay the inevitable. Ironically, there is still the potential for shortages against delivery, as demand patterns from the end-customer grow more volatile. How far does this situation have to go on before products are requalified against different production line configurations with different rates that can better meet modified demand flows? The resistance to change in such circumstances is high because requalification of product for a different configuration takes valuable engineering time, which is now busy taking care of the next wave of products coming in. It leaves the planning team at a loss, as well as the factory.

Meeting the delivery need. The Lean approach is to permit the target line for the product to be determined through the optimization of the plan that meets the delivery need, and to be free to change it over time. In today’s most advanced production plan system, this also includes the determination of which products are put together to form the groups for common material setups. The change in the data preparation flow to permit this to happen is to qualify all incoming products against all possible production configurations, so as to provide the optimization engine of the production planning the freedom to use any of the capable processes.

Good production planning tools will be able to simulate the line throughput rates for each product as part of the planning optimization itself, so the engineers at the data preparation stage do not need to make the actual machine programs. Qualification of products against multiple line configurations is relatively easy to do when using a multi-vendor data preparation and programming tool because a single central product model can easily be tested against each line configuration, which can consist of any vendor’s machine. When remaining dependent on the simple machine vendor-supplied tools, this would become a major pain because data translation, especially of material shape libraries, has to be done manually and repeated for each configuration.

Therefore, the flow for engineering is changed when the Lean flow is adopted. Rather than data preparation, line assignment, SMT optimization, and then planning optimization, the flow becomes data preparation, product assignment qualification, planning optimization, and then optimized SMT program generation.

This process does involve investment in tools over and above those provided by machine vendors, which can still be used for the final machine program optimization, but the rewards can be significant.

New machine investment far exceeds the investment in software that would have a similar effect of improving throughput levels. The difference also is that the application of Lean to the engineering flow removes waste. Investing in more machines before investing in the right software tools simply amplifies the existing wasteful processes, consuming investment as well as shop-floor space and additional strain on already over-stressed resources.

Michael Ford is marketing development manager, Mentor Graphics (mentor.com); michael_ford@mentor.com. His column runs bimonthly.