Skip to content

Progress made in PV production, more is needed

April 16, 2009

 In 2007 at the PV Fab Managers Forum in Dresden, Germany, we heard about the “importance of improvements in equipment and manufacturing to bringing down the costs of solar PV in the future” http://tinyurl.com/c2yck6  In an article on the 2008 PV Fab Managers Forum “Photovoltaic Fab Managers Forum Review”, by Mark Osborne, http://tinyurl.com/d3zjdj , the conclusion is a call for more fab improvement. He concludes “While current demand outstrips supply for PV modules, [how fast things change], it was made clear by attendees that unless cost-per-watt and volume production goals are achieved in the short and medium range term, the renewable energies marketplace could lose patience with PV as a viable mainstream energy generator” The 2009 Forum continued that theme with an emphasis on supply chain improvements. In “PV Fab Manager Wish List: Achieving an Efficient Supply Chain”, http://tinyurl.com/c63h4b , by Eddy Blokken, called for improvement in manufacturing and cost cutting. I think the consistency of the perceived needs proves their validity. Proven tools are needed to help solve these problems. An agenda for improvement in a  PV fab might consist of  Equipment utilization/Overall Equipment Effectiveness, (OEE is availability x performance x quality) , equipment material state model, enable rerouting of material, dispatching, and advanced planning and scheduling, http://tinyurl.com/d2dmjg . This agenda can be tackled with a good APS tool like ManSim.  With a validated ManSim model rerouting products around down tools, dispatching, order planning can all be addressed. What-if analysis can be done to find the best solution to a problem, before you assign scarce resources to implement a change. Free white papers are available for download on this blog site. As always, we like to hear your comments.

Modeling KAIZEN

April 8, 2009

From Wikipedia, “When applied to the workplace, Kaizen activities continually improve all functions of a business, from manufacturing to management and from the CEO to the assembly line workers”. Having a model of the entire manufacturing system gives focus to projects and a safe way to test ideas. Modeling can have many benefits. Some of them are:

   Building and validating an enterprise model gathers operators, managers, and process owner together. Consensus can be built around a validate model.

   The model is a focus over time; keep pace with continuous improvements and fostering new ones.

   The validated model is an excellent map of existing processes and illustration of how they work.

    The model can serve as test bed for change at one workstation, while simultaneously showing its effect on the organization as a whole. This micro and macro predicted results’ reporting shows if the localized treatment might kill the patient.

   A validated model can educate and involve everyone. Whether the suggested improvement came from a station operator or Operations manager, the predicted effect is visible to all. The operator may see her suggested improvement get more units out of her station in the model, but fail because the vendor can’t supply the needed parts fast enough two stations downstream. The manager who wants more out by simply putting more in to the factory will see WIP rise and cycle times increase.

   By focusing on the model of the system everyone can learn to think of manufacturing as a system. Localized thinking and slowly be replaced with global thinking. Organizational benefit will become the key criteria for evaluating a suggested improvement.

   Knowledge of the enterprise as a system will spread up and down the organization. Managers will see the effects of ideas on operators and operators will learn how they affect the enterprise far beyond their stations.

With an enterprise model to test lean improvement ideas all employees’ knowledge will be improved. Buy-in will be easier to achieve.

Your comments are welcome.

Free downloads are available on this blog site.

 

Can You Accurately Plan Orders Based on Spreadsheet Models?

April 2, 2009

What is the first thing a planner has to know when planning orders? A typical answer is expected cycle time. Where does that answer come from? How reliable is it?  Is it based on history collected from a specific set of circumstances that no longer exist? Or is it dynamically calculated, looking forward predicting results of the constant changes taking place in the factory? Many factories still employ a spreadsheet to determine predicted outs cycle time for products. Depending on spreadsheet predictions of cycle time may lead to problems of overloading the factory, poor serviceability, longer than predicted cycle times, over and underutilized equipment. Some of the causes of these static model problems are:

1.      Using average historical cycle time at each step, products can move independently of each other. This allows for process of more than one lot a time where it is not feasible.

2.      Using historical static calculations of moves, setups, queues, maintenance, reflects past product volumes, mixes and factory behavior. This can lead to over or under estimation of cycle time in a new set of circumstances

3.      Typically these models use only first in first out sequencing for lot movements. This sequence remains the same from start to finish of processing. This is unrealistic and leads to skewed results.

4.      Spreadsheets typically can’t account for all the variation in the factory, such as queuing at down equipment. This can lead to overly optimistic cycle time predictions.

One way to model a factory, taking into account variance is with discrete event simulation. These models can take all the variability of mix changes, unpredictable equipment, setups, different product priorities, WIP levels and calculate a cycle time based on dependence on all these factors. Using this as a basis for planning orders allows the factory to operate in a much more predictable manner. This results in better on time delivery, lower cycle time and meeting key measures.

A detailed white paper titled “Using Expected Cycle time Values to Manage Change” is available as a free download on this blog site. It details the difference between spreadsheet and discrete event simulations models.

We’re very interested in your comments. You can make them on this blog site.

Semiconductor, films of natural phenomena, beautifully rendered

March 30, 2009

British artists Ruth Jarman and Joe Gerhardt have created a series of beautiful films about magnetic fields, solar winds and scientific concepts. They call themselves Semiconductor. Visit their site and select the Artworks tab. Under that tab are thirteen five minute films. http://www.semiconductorfilms.com/ 

Can outs be pumped up and cycle time shortened in existing fabs?

March 24, 2009

Several publications recently had articles considering ways of getting more wafers out of existing fabs faster. I thought in this period of low WIP and over capacity, cycle time is down. Why worry about it now? But, maybe this is the time to look at the existing fab and consider ways to improve throughput and cycle time, before the next big surge in orders hits and there isn’t  time. What are some of the methods, procedures that could be improved? Each fab will have their own ‘special’ issues. But, in general, here’s a few areas many fabs have issues in:

1.      Lot Size, especially <25 wafers

2.      Dispatch rules.

3.      Predictive PM schedules

4.      First wafer delay

5.      Supply chain

6.      Lot scheduling

7.      WIP

8.      Cycle time

9.      Set up time

10.  AMHS

11.  New process introduction

We can’t investigate all these points at once, but we must consider them all together, as they interact as a system. One of the best ways to do this is with simulation. ISMI has teams of engineers modeling current and future fabs in an effort to find the best ways improve productivity and cycle time, http://tinyurl.com/d5fsg9 . Rank your issues highest pay back to lowest and start at the top of list. With a validated model of your fab you can start attacking  one issue at time. Using what-if analysis you can model every option you can think of until you get most practical, cost effective solution. There are more suggestions and papers to download at  this blog site

 

 

 

 

APS -Why is the implementation decision so hard?

March 21, 2009

What is really hard to do is to get people to move to APS (simulate the factory when planning). The facts are that management relies on averages and history to run their factories. Weekly and monthly meetings for factory status usually create and use averages, static calculations, and history for current status and future predictions. This approach diminishes a factory’s ability to react to current factory status and its future behavior. History is important to report and learn from, but its recurrence in the same detail is rare. The results are plans that lead to sub-par factory performance and chaotic situations (expediting, overtime, additional costs). Understanding the detail of constraint areas and using APS’s discrete event simulation for prediction will move a factory from average performance to excellence. This always seems to overwhelm management (“We don’t have people or time to do that!”). A complete analysis of the factory to create the simulation model is not needed. A focus on understanding and analyzing the known factory constraints will yield big improvements in planning.

History and averages that are typically used in MRP, ERP, or spreadsheets degrade the ability to estimate the future. In a less complex factory, it may give you 75% accuracy. In a complex industry such as semiconductors, the accuracy is much lower. This seems to be good enough for most management in the USA (“We can manage the factory this way.”). Convincing people that their plans are only at 75% accuracy (or worse) and getting them to apply manpower to improve by implementing APS is very HARD to do. Why don’t we seem to be trying to get closer to 100%? Is it because the competition is equally lethargic? Is it we are too timid to change the current planning system? Is it organizational resistance? What do you think?

Sustainable Lean Manufacturing

March 20, 2009

The toughest nut to crack in putting Lean Manufacturing’s excellent ideas into practice is executing organizational change and sustaining that change. After the initial mind-set change in the organization, the sound principles of Lean Manufacturing are implemented and cemented in place by a set of Lean Manufacturing run rules and procedures. The factory is lean. But, wait. It is two years later and we have new personnel with the traditional business-as-usual agenda in charge. The factory gets fat again by overloading the factory in the quest for revenue, expediting parts as serviceability slides downward, inflating cycle time, worshipping throughput, and thus being forced to use frequent overtime. Just like food diets for people, weight is lost and then gained right back again. I have seen this at many factories where cycle time was cut by at least 50% only to regress to the old performance a few years later. What about all those Lean run rules and practices? How can we stop them from fading? One thought is to use a factory simulation tool to capture and maintain these run rules and procedures. Make it the centerpiece of the Lean Manufacturing structure before, during, and after the Lean Manufacturing effort. When the business-as-usual crowd comes in and forces their so-called “operational improvements”, crank up the model and show all the stake holders the damage of Fat Manufacturing. Keep that slim factory shape forever and improve it as time goes by.