•  Fresh outside perspectives

•  Expertise

•  Unbridled firepower, since they’re not beholden to the daily production grind

But bringing in a consultant, if done wrongly, can be disruptive—and even destructive—to a firm.

The wrong way

I experienced the wrong way first-hand when I was employed at a company as its sole industrial engineer, many years ago.  I had a list about ten pages long of all the opportunities for improvement that I had identified in its manufacturing, ranked by ROI (return on investment).  I made rapid progress completing projects and knocking them off the list, but it was disheartening seeing all the money that we were still leaving on the table, being that I was only one guy.  Try as I may, my boss would not or could not hire more engineers.

One day, I heard that the vice president of operations had had a consulting firm come in for $30,000 for a couple weeks to assess opportunities for improvement in our manufacturing.  I saw the document that the firm produced and almost all of their ideas were ones that I already had on my list.  I showed the V. P. my list and he was surprised.  I was dismayed that my company had seen fit to have an outside firm come in to do what I had been chartered to do, without even consulting with me.  If done properly, I would have welcomed the consultants, appreciating the long-overdue extra manpower and enjoying working with them as a team, showing them what I had, and putting our heads together to identify even more opportunities for savings and working together to execute our plans.

That would have been the right  way.

But, as it was, I firsthand experienced the wrong  way to bring in consultants.

The right way—3 rules

1. Bring in a consultant who knows that  the front-line people who’ve worked in a company for years have more ideas and insights than even the smartest outsider.
2. Bring in a consultant who knows that his job is not to displace local experts, but to employ them as a foundation to create-—together-—even more ideas, then work together with them to implement them.
3. Bring in a consultant who knows that working as a team with local experts—as equal partners—not only brings the tangible benefit of making use of all the employees’ experience and ideas, but it also wins their psychological support.  When they see that the consultant will be an ally and source of empowerment for them—not an opponent—they support him fully.

The post 3 Rules for Getting the Most Out of Operations-Improvement Consultants appeared first on Simitar Operations-Improvement Consulting.

The post How “Overstaffing” at Bottleneck Machines Can Unleash Extra Capacity appeared first on Simitar Operations-Improvement Consulting.

The simplest—and highest-profit-margin—modeling project that Simitar founder Bob Kotcher has ever done demonstrates the power of simulation modeling for operations improvement.

Celebrate your inner Seinfeld

At a restaurant one day, my inner Seinfeld came out (we all have one—come on).

Since the Americans With Disabilities Act was passed in 1990, many restaurants had to remodel their restrooms to make them wheelchair accessible. This often required them to remove internal partitions that comprised the stalls. With privacy gone, they put locks on the restroom doors and made each restroom single-user.

BUT…many businesses failed to remove the male or female signs from the doors. Furthermore, even some new restaurants, building new single-user restrooms from scratch, put male and female signs on the respective doors. Why not put unisex signs on each door? We now often have the problem of going to such restrooms, seeing a line in front of “ours,” and having to wait (and wait, and wait, and wait…) while the other restroom sits enticingly empty (as our meal buddy out on the table grows increasingly bored).

This calls for an engineering analysis…

Suppose we did a time study on the restrooms (yes, I am an engineer) and used the results to make a static capacity model in a spreadsheet? We’d almost certainly find that each restroom is loaded far below capacity—maybe 75% at peak hours. If we made them unisex, average loading would remain the same. So what’s the problem with them being dedicated by gender?

Well, a computer simulation model would show how, by replacing the two restrooms’ signs with unisex signs, average queue time would decrease.  This is because, due to random variation, there are times when there are women in line when the men’s room is wide open, and vice versa.  Making the restrooms unisex would reduce waiting time in such situations.  Reduced waiting would perhaps result in more satisfied customers and more repeat customers—all for about $20 for new signs.

From restaurant to wafer fab (they both offer chips)

Now, I doubt that restaurants will be conducting simulation analyses to see if spending $20 for new signs would be a profitable investment or not. But one internal client at a wafer fab did something similar.  You see, in wafer fabs and other factories, there are often several identical machines processing in parallel at a particular operation.  If they’re running different recipes, process engineers often like to dedicate each machine to a particular recipe.  This makes process control easier.  But it increases average waiting time due to the above phenomenon.  What is the best tradeoff?

At this client, a process engineer oversaw a wet bench with two parallel identical baths, the only difference being that they were set at different temperatures (dedicated by gender, if you will). Recipe A required temperature A, and all other recipes required temperature B.  The engineer found that he could run the Recipe A wafers at the B temperature if he made Recipe A’s processing time longer.  This would reduce average queue time for all wafers…but increase the processing time for Recipe A wafers.  His question to me was: if I set both baths to the same temperature, will the reduced queue time for all wafers outweigh the increased processing time for Recipe A wafers?

I pretty quickly did a simulation analysis and found that, yes, setting both baths to temperature B and increasing Recipe A’s processing time would actually reduce  average cycle time.

Act on the analysis results, bank the savings

Armed with the analysis results, the engineer made the change. And the result was the biggest return on investment of any simulation project I’d ever done! Not because the savings were so mammoth, but because the cost of the rather simple analysis plus the cost of making the ensuing change were so low. But the change never would have been made had the simulation model not been available to test it out, because nobody in his right mind would approve a change that would increase loading on a machine in order to reduce  cycle time. Until a simulation analysis showed that it did.

The post Restroom and Wet-Bench Equality, Now! appeared first on Simitar Operations-Improvement Consulting.

Dynamic 

Dynamic capacity models account for the variability of the real world

In operations with variability, executives know that they cannot run them close to 100% loading because queue times become unacceptably high. They know to invest in surplus capacity—not to increase throughput, but to keep queue times in check.

But what is the optimal amount of surplus capacity to purchase, and exactly where? What is the optimal capital-equipment (capex) purchase plan?

The high cost of static capacity modeling

Many companies use static spreadsheet models for this and swag a target loading number: if loading goes above X percent at any operation, they invest in more capacity at that operation. In wafer fabs and fab-like operations, I’ve seen X as low as 65% and as high as 95%.

But swagging a number like this leaves massive amounts of money on the table—$10 million+ a year for a typical wafer fab. How?

Because one is investing in some equipment that actually doesn’t do much to reduce cycle time per dollar invested, and not investing in other equipment that actually reduces cycle time quite a lot per dollar invested.

What a dynamic model does that a static model doesn’t

Conducting a dynamic capacity analysis enables one to see the correlation between loading and queue-time contribution for each machine type. Then one can invest in machines in order of cycle-time-reduction per dollar. This way one can achieve the same throughput and cycle time of the spreadsheet method, but often for millions of dollars less. And what’s great is that the savings are all quantifiable—one simply compares the machine set recommended by the spreadsheet model with the machine set that the dynamic model says will produce the same throughput and cycle time. Then look at the price difference. The capex savings in the first year is often 10-20 times the cost of building the model. The return on investment for a dynamic capacity modeling project is thus easily quantifiable and justifiable.

How is a dynamic model built and operated?

How, exactly, is dynamic capacity modeling done? A dynamic model incorporates the same information that goes into a spreadsheet model-—and a bit more. The data is put into dedicated modeling software. Upon the “Run” command, an internal clock starts in the software, products start in the model, machines start breaking down and getting repaired, parts start occasionally requiring rework or are scrapped, operators start going on break or calling in sick—-all the pleasures of the real world. All of the above occur randomly, but in accordance with their real-world probabilities. When the run is finished, we now have a critical new statistic: time. We can see what every product’s queue time was in front of each machine. Rerunning the model with different quantities of machines, we can see the correlation between various machine-purchase options and cycle time.

Types of operations that are best candidates for dynamic modeling

Any operation that is variable and/or complex. Great examples are wafer fabs, job shops, hospitals, and many supply chains. Few operations are so simple and determinate that they cannot benefit from dynamic capacity modeling.

The post Save Up to Millions of Dollars a Year in Capex: Use Dynamic Capacity Planning appeared first on Simitar Operations-Improvement Consulting.

What is the best method of capacity planning

For a deeper dive on this subject, see this Simitar article, which was published in the FabTime newsletter (see page 5):

FabTimeNewsletter13 02

[This issue of the FabTime newsletter is made available with express permission of FabTime.  To subscribe to receive future issues of the FabTime newsletter, please visit FabTime’s website at http://www.FabTime.com/newsletter.shtml.]

The post What Level of Capacity Planning is Right for My Facility? appeared first on Simitar Operations-Improvement Consulting.

Overview of modeling

Unlike most other operations-improvement consultancies, Simitar is expert in computer-simulation modeling. 

When operations are highly complex or variable, computer-simulation modeling can reveal huge opportunities for savings that spreadsheet models—and even experienced observers—overlook.  That’s because simulation models take into account the variability present in real life.  Simulation models also include a critical factor that spreadsheet models ignore: cycle time, and how it varies with load. 

Once optimal parameters are found in the simulation model, they can be applied to the real operation.  The result can be significant improvement in all aspects of operations performance, as well as reduced capital-equipment expenditures. 

Any type of operation can be modeled

Essentially any manufacturing, service, or business process can be modeled. Simitar personnel have modeled:

• Entire wafer fabs containing up to 1500 production steps and 170 types of production equipment spanning multiple sites around the world.
• Mid-sized production areas to assess complex interactions between machines and to estimate profit-maximizing staffing levels.
• Individual pieces of production equipment for internal throughput optimization.

Service industries and business processes are heavy beneficiaries of simulation modeling. Examples are hospitals, distribution networks, rail systems, airports, call centers, and claims processors.

A model can help you make thousands of decisions

People unfamiliar with simulation are usually not even aware of the vast variety of questions that a simulation model can answer—they don’t know which questions to ask. Here are a few examples:

• What capital equipment should I order to meet next year’s throughput and cycle-time goals at minimal cost?
• What combination of dispatching rules, setup rules, batching rules, and operator allocations will most improve my throughput, cycle time, on-time delivery, and cost?
• Where are WIP bubbles likely to form during the next shift, and what can I do to preempt them?

One-time or permanent model?

Models can be built for a one-time decision or can be maintained and used on an ongoing basis for continuous improvement and ongoing capacity planning.

The post How to Improve Factory Performance via Computer-Simulation Modeling appeared first on Simitar Operations-Improvement Consulting.

Here is a graphic with instructions on how to construct the tool:

The full paper on this method follows.

Abstract

This paper describes a method for using Monte-Carlo simulation to estimate the profit-maximizing loading for each machine in a factory.  Multiple runs of a factory simulation model are made—with varying machine quantities—with the results graphically displayed as a correlation between static % loading and ROI (return on investment) for each machine purchase option.  The ROI is calculated from the estimated dollar value of the cycle-time reduction expected from that machine purchase and the purchase cost of the machine.  Management can thus graphically see the ROI for machine-purchase options for a variety of machines and compare them.  Such ROI-loading graphs—or “ROIL” graphs—once created, enable static models to benefit from past simulation runs: users graphically see the target static % loading of each machine type and also see how it will change if the budget or the required ROI change in the future.

1        OVERVIEW OF THE TRADITIONAL METHOD: STATIC-CAPACITY ANALYSIS VIA SPREADSHEET

The most basic way to model capacity is via a static spreadsheet.  A static spreadsheet takes into account machine quantity, machine downtime, number of visits per product type, processing time per visit, and other factors to produce a static % loading for each machine type.  When such a spreadsheet predicts that a machine type will be loaded more than a certain percentage in the future, action is  taken—such as ordering an additional machine—to reduce loading back below this number.

1.1         Shortcomings of Static-Capacity Analysis

Static capacity analysis’s major shortcoming is that it does not capture the effect of variability, which can be very high in some operations, such as wafer fabs.  Wafer fabs often must run at the very edge of technological capability to produce products which the market demands.  Machines thus pressed to the limit often suffer high downtimes and high rework and scrap rates.  Furthermore, in wafer fabs, machines are often visited many times by the same wafer—even approaching 100 visits—meaning that a neat production-line layout is not possible.  As a result, wafers must be regularly transported around the fab, incurring variable delays.  In many fabs, the machines are also used concurrently by R&D.  Combine this with up to 1500 operations required per completed wafer, and many product types running in the same factory, and you have more variability than with just about any other type of manufacturing operation.

Any operation with such high variability, if it were run at near 100% loading, would see exorbitant cycle times.  And low cycle time is especially important to fabs running R&D on the same machines because quick cycle time speeds up R&D learning cycles, meaning new products get developed and into the market more quickly, while price premiums are at their highest.  Mature products also gain an advantage from quick cycle times, given that technological advancement is reducing their market value daily.   Furthermore, short cycle times enable agile response to changes in customer demand.

Because of the variability inherent in wafer fabrication, and the high cost of cycle time, a fab typically sets its maximum allowable loading at some level well below 100%–65-85% is typical.  This is considered to provide the most profitable tradeoff between the cost of equipment and the cost of cycle time.  In choosing its maximum static % loading then, a company is choosing a de facto maximum acceptable cycle time.  The capacity of a fab, given such a self-imposed “ceiling” on cycle time, has been coined “cycle-time-constrained capacity” (Fowler and Robinson 1995).  This is the capacity of a fab, given that it will not accept a cycle time higher than a certain value.

So, if a fab’s maximum allowed static % loading is, say, 80%, a machine type that is forecast to be loaded 81%+  in the future would have that mitigated by a machine purchase or other action, whereas a machine type forecast to be loaded 80% or less would be left as-is.  But here we come to a major shortcoming of static capacity analysis:

Relative to static % loading, not all machines contribute to cycle time equally.

Furthermore:

Not all machines cost the same.

A simulation analysis could enable you to identify those machines that contribute the most to cycle time relative to static load.  Adding machine cost into this equation enables you to calculate a  cycle-time reduction per dollar for every machine-purchase option (Grewal et al [1998]).  Focusing your spending on your purchase options that produce the greatest cycle-time reduction per dollar enables you to achieve your cycle-time goal at lower cost—perhaps millions of dollars lower cost.

For example, let’s say that your static model projects that, in a future quarter, Machine A will be loaded 90% and Machine B will be loaded 70%.  Using a traditional straight-across 80% cutoff, you would buy another Machine A and not buy another Machine B.  But in reality, buying a Machine B may actually bring you greater cycle-time reduction than Machine A.  Why is this?

1.2         Some Factors That Cause a Machine’s Cycle-Time Contribution To Be High Relative to its Static % Loading

• High number of visits per wafer.  Take two identical machines with the same % loading and the same average queue time per visit.  But if one is visited once per wafer, and the other is visited fifty times by each wafer, the latter’s total queue time from this machine will be fifty times that of the first machine.
• Coarse machine granularity.  All else being equal, a single large-capacity machine will contribute more to queue time than a number of small-capacity machines.  For example, take two machine families, identical except that one consists of ten machines, each having a capacity of 50 wafers a day, and the second consists of a single machine, having a capacity of 500 wafers a day.  For the latter machine, capacity toggles between 100% and 0% as the machine goes up and down, creating queue-time spikes followed by lost production.   The ten small-capacity machines, on the other hand, won’t see such drastic change in capacity and will provide a smoother flow of product to downstream machines, resulting in shorter average queue times.
• Batch machine.  All else being equal, a batch machine has higher queue time relative to % loading than a single-wafer processor.
• Greater variability in downtime.  Take two identical machines with identical downtime percentages, the only difference being that the second machine fails less frequently but for longer periods of time.  The second machine will have longer average queue time than the first.

1.3         Enter The Simulation Model

A simulation model can take into account the above soup of factors and others and tell us the correlation between static % loading and resulting cycle-time reduction for each machine-purchase option.  At Simitar we realized that we could dollarize the benefit of this cycle-time reduction and compare it to machine cost to calculate an ROI  (return on investment) for each machine-purchase option.  Graph all of these machine-purchase options, and you have what Simitar calls a ROIL graph.

2        THE ROIL GRAPH

Figure 1 shows a ROIL (Return On Investment-Loading)  graph created with a simulation model.  You’ll notice that the ROIL graph correlates % loading—on the horizontal axis—with ROI (return on investment)—on the vertical axis.  This is simply a standard “operating curve”—which correlates % loading with queue time—but with queue time dollarized  and combined with the purchase cost of the machine to enable replacement of queue time with ROI.  Details on how that is done is described in section 2.1.

Notice that there is one curve for each machine type.  At a particular current % loading, the indicated ROI is what you’re likely to see if you purchase an additional machine of that type.  For example, let’s say you currently have three Machine A’s—loaded 64% in a static model, per the Figure-1 graph—and nine Machine C’s—loaded 80% in a static model, per Figure 1.  Purchasing an additional machine of which type is the more profitable investment?  Using static analysis, it is obviously the more heavily loaded machine type: Machine C.  But the ROIL graph shows that the three-Machine-A point is higher on the ROI scale than the nine-Machine-C point, so purchasing another Machine A is the better investment—it buys more cycle-time-reduction per dollar.  Purchasing a Machine A yields an ROI of about 95%, vs. only about 45% for a Machine C.

Figure 1: the ROIL graph shows the expected return on investment for each machine-purchase option

Another way of looking at it is by following along one curve.  For Machine A, if you have just two machines, the ROIL graph shows that they’re loaded 96%, and purchasing an additional one yields an ROI of 490%.  If you have three Machine A’s, the ROIL graph shows that they’re loaded 64%, and buying an additional one yields an ROI of 95%.  If you have four Machine A’s, loading is 48% and the ROI of purchasing an additional machine is 25%.  You can see how, as a machine type becomes less loaded, you see a lower ROI for each additional purchase.

2.1         Creating a ROIL Graph

Ranking purchase options by cycle-time-reduction-per-dollar is not new.  Creation of a ROIL graph takes this further by (1) incorporating the dollarized value of cycle-time improvement which, when combined with machine purchase cost, enables calculation of ROI, (2) graphically showing the result of multiple machine-purchase options, and (3) showing how the profit-maximizing machine set will change if the budget or required ROI change.

The post Simitar’s Handy Capacity Reference Tool for Managers appeared first on Simitar Operations-Improvement Consulting.