Moshe BenBassat
Chairman & Founder, Plataine

Introduction
Facing a global economic crisis, Metal Fabricators must cope today with an ever changing production environment, subject to the volatility of the current economic conditions and increased pressures for higher quality, faster turnaround time and reduced pricing. While in many cases production volume is down, the volatility of sales is going up with orders made at short notice, and then reduced, expanded or cancelled altogether. In addition, the growing trend of high-mix / low-volume orders with ever-shortening delivery times, is adding further complexity to the already challenging production floor environment.
Under these circumstances, managers find themselves spending a great amount of time planning and re-planning their production schedule only to make more changes facing everyday interruptions like machine breakdown, material shortage and others. Meeting this challenge takes great expertise, experience and a strong handle on current status.

Conversely, failing to meet it costs time, money and customer dissatisfaction in an era where customers are valuable than ever and every dollar counts.
In many cases, modern software products for production floor optimization alleviate a significant part of these complexities, helping plant management achieve better customer service, higher material utilization, lower level of inventories, reduced costs and overall higher productivity.

 

The Production Reality
Today’s production environment calls for continuous monitoring and dozens of daily decisions that impact the plant’s both top and bottom lines. These decisions often represent tradeoffs between multiple often conflicting objectives. Making the right decision at the right time - every time - requires not only considerable experience and decision making skills, but also accurate data in real-time.

Unfortunately, in many cases these decisions are based on the wrong information or assumptions, and are often made by employees who are not aware of the big picture and whose decisions significantly impact customer satisfaction and financial profitability. In our meetings with plant managers and owners, we often come across the following examples for such ad-hoc decisions:

· Example 1: Sales decides that a given order is urgent and requests a break-in to the schedule. Accordingly, the production manager gives this order a higher priority and pushes scheduled orders forward, resulting in bottlenecks and additional costs. Profit or loss on this specific job is realized at month end closing, rather than before the job runs - when decisions can be made to pass accurate incremental costs along to the customer.


· Example 2: Uncertainty creates ‘Process just in case’ WIP and excess finished goods inventory: an order is processed “right away” at the entry equipment (e.g. laser, punch, or plasma table), but is then moved to WIP where it spends hours or days before moving to the next stage.


· Example 3: Raw material or WIP is late arriving to equipment. Operators and line supervisors must make decisions on the fly upon this realization. Time is taken away from other tasks and poor choices are often made in the attempt to prevent the impending bottlenecks downstream.

These examples are symptoms to a more systemic problem, one that has the following root causes:

1. Insufficiently useful Data: True real-time availability of accurate data, presented in a useful format, in support of good decision making.

2. Lack of clarity across the organization regarding management policy for job priorities, profitability, customer service, etc. These may be in conflict one with another, and setting the ‘right” balance for any given company is a strategic decision. Without such a policy, or without the ability to act on it, decisions are sporadic, inconsistent and yield painful results.

3. Lack of decision support: The Production Floor environment is intense and complex. Arriving at good decisions - let alone optimal - involves the analysis of many factors, complex calculations, and comparing numerous alternatives which are well beyond the capabilities and speed of any human brain.

“I’ve Got Pat…” and capturing tribal knowledge
So, how is it done today? It’s obviously working, yet in many cases there is a lot of room for improvement.
In a recent visit to a manufacturing plant I held a discussion with the plant manager about his daily operations. Several business scenarios were discussed, from building the schedule for the next day and week, to responding to unexpected events. When asked how things are done at his plant, a frequent answer by the plant manager was: “I don’t know, but I’ve got Pat”.

 

‘Pat’, representing the employee that ‘knows all’ and that ‘handles all’, encompasses within her years of experience and knowledge of corporate polices and priorities. The “Pats” of the world will agree their work is a challenge, but they are not always able to articulate what is driving their decisions. Moreover, their decisions are typically based on limited “good” data and very minimal decision support systems. Moreover, when Pat is not around, the quality of decisions quickly deteriorates.

 

Pat, as successful as she may be, is facing the basic limitations inherent to us all: a limited ability to simultaneously consider numerous factors and data items to identify the “best” solution from a large number of legitimate ones.
Computerized solutions that collect information, and that are able to act on it in real time can be of great assistance to Pat and her Plant Manager. They embed in them the mechanism for collecting raw data and transforming it to relevant information, and can then automatically act on it, based on preset policies and priorities as defined by management. The true value of such systems is their ability to quickly generate thousands of alternatives, and then pick the best one. This, of course, is prohibitive for the human mind.

Optimization software could therefore be Pat’s “best friend”. It will encapsulate her knowledge, and not only serve as the “watch dog” for potential mistakes, but will also automatically produce the right answer at any given situation. With such a tool at her disposal, Pat will be able to focus her attention on exception handling, dealing only with situations that are beyond the boundaries of the routine policy. Lastly, when Pat is off duty, nothing will represent her better than her best friend and the production floor will continue to flow properly until Pat returns.

 

Production Floor Optimization
Optimized Metal Fabrication plans are aimed at balancing multiple and often conflicting objectives stemming from multiple sources: shorter delivery times, volatile raw material cost, cheap off-shore labor, and the ultimate need for growth and profitability.

Searching for the “best” plan means that we are able to compare any two plans and judge one to be better than the other. Such an evaluation function - also known as objective function - typically combines several factors, including material utilization, cutting & assembly time, customer satisfaction and others. Whatever the objective function is, there are some principles that are likely to improve the outcome of the “best” plan.

· Principle #1: Increase flexibility and degrees of freedom, for example:
Higher flexibility in mixing parts of different products or customer orders that use the same raw material yields better material utilization for the optimal plan. The basis for this principle is the classic “Rocks, Stones, Sand” principle. Namely, for a given sheet of metal, the higher the variety of parts which can be considered for placement, the higher the potential utilization. If you only have large parts to nest, the opportunity for optimizing material utilization is very limited.

More on this in the example below.

· Principle #2: The longer the planning horizon, the better overall performance you can achieve. The further ahead you look, the more information you have for customer demand and the higher the number of options you can consider for optimization.

 

· Principle #3: Continuous optimization enables overall better performance. What was optimal an hour ago may not be optimal any longer. At any given time, new information comes in: jobs take longer than planned, new high priority jobs come in, a machine may be down, raw material is missing, etc. Being able to act in real-time and modify the plan to continually maintain optimality ensures that we squeeze the most out of the available capacity and respond in a manner which is consistent with our business priorities.

· Principle #4: Comprehensive, simultaneous coverage of all variables produces optimal solutions with better values. For example, while optimizing the nest for material utilization, you cannot ignore the cutting time or the tool path. The optimization process must consider both of them simultaneously.

Production Business Rules
The essence of producing an optimal production plan is searching for the “best” plan among a set of “legitimate” ones. The search process itself revolves around a systematic and efficient exploration of various combinations. However, to come up with a practical production plan, we must limit our search to those “legitimate” plans that are consistent with the business and logistic constraints of running a production floor. These aspects, of course, may vary from one company to another.

A legitimate production plan is defined by a set of must and must-not rules, grouped into two categories:

· The first type of rules is dictated by technological and logistic considerations which - if violated - will yield unacceptable results. For example, consider a rule that says “In precision Laser-Cutting of metals, when nesting multiple small parts together, make sure they are properly tabbed“. The reason for this rule is that as we cut small adjacent parts, some may flip, potentially crashing the laser-head or generating scrap. Such rules are an inherent part of the process and do not leave much room for flexibility.

· The second type of rules is self-imposed, due to business and operational practices that may or may not be truly necessary. For example, rules such as “Do not mix parts from different orders” or “Product X is always processed on machine A”. It is important to understand that as the number of rules increases, the ability to optimize our plan decreases, as more constraints limit our degrees of freedom. Therefore the business implications of every rule should be carefully evaluated and revisited periodically.

Reducing the level of rigidity of certain rules may ultimately lead to significant improvements in the outcome of the optimal plan because it increases the degrees of freedom for decision making. This does not necessarily mean that these rules will be totally ignored, but rather applied selectively according to one’s business policies. The example below demonstrates how relaxing a common production rule yields significant material savings. In this case, rather than broadly applying the rule “Do not mix parts from different orders”, we could use a less restrictive version such as “It’s OK to mix parts of different orders, but within no more than a 2 sheet distance between parts of the same order.”

 

When formulating rules, one needs to ensure they capture the desired outcome. However, in tight situations, or when an opportunity arises one may wish to consider rule relaxation: “I’ve got a business to run, and I’m OK violating [some of] my own rules”

A policy for expanding circles of rule relaxation could be very powerful to handle tight situations or seize opportunities to generate business gains. Once such a policy is created, a computerized system can practice it as a matter of routine, and surface any exceptions along with the appropriate data for the final call by Pat or the attending supervisor. Without such a system, manufactures that are pressed to get the job done may often overlook such opportunities for improvement, let alone have the time to reshuffle the order and provide an optimal solution. In such cases, modern optimization software not only highlights the opportunity for improvement, but also provides the optimal solution that takes advantage of it.

 

Figure 1 demonstrates the standard practice today where nesting and cutting programs are generated ahead of time for certain kits/products and are stored in the production library. When an order arrives for a certain number of units, the library program is retrieved and replicated as illustrated in Figure 1.

In Figure 2 we show the material savings achieved using modern use of real time nesting and cutting optimization software that allow for a business rule that enables mixing of parts. The end result is a 25% savings in material consumption, utilizing 3 sheets compared with 4 for producing the same parts.

Figure 1: conventional (“Static”) nesting requires 4 metal sheets

Figure 2: optimized nesting, combining orders of the same material, requires only 3 sheets

 

The right Mix of Man and Machine
True optimization software must stand alongside the user. It does not only serve as the “watch dog” for potential mistakes, it also automatically produces the right answer at any given situation. The basic idea would then be to share responsibilities between partners (man, computer) where each partner does what it is good at: Let the shop floor managers describe the business and the rules of the game, and let management spell out the company goals and other policy directions. But, at the same time, let the computer practice this policy in a consistent and fast manner while producing optimal decisions subject to the rules of the business. In this context, Pat’s role will then be focused on dealing with exceptions, where her experience really matters: she will no longer spend valuable time developing and evaluating alternatives because the computer will most likely do it better and faster.

Devising production plans that are both optimal and practical is an effort that requires a combination of process, technology, and people. Leading manufacturers are adopting new business practices and technologies that embed the optimization concepts described here with encouraging measurable results.

When seeking new technologies to help us optimize production, we need to look for those technologies that enable new processes and drive considerable productivity improvements and cost savings. Modern software technologies that include automatic decision support and optimization open new options to improve factory floor operations. Today, there is a growing trend of companies that are adopting such technologies and enjoying the benefits it can offer. In future papers we will elaborate more on optimizing the production plan to a given objective and how to consistently execute optimal plans in the plant’s dynamic environments.

 

To download a PDF version of this article - Please fill in the form below