Latest Blog Entries
Archives

08 Oct Why Bother With An Optimizer

(By Mike Raftery)

There seem to be two main camps on the topic. Group one thinks there’s a problem that exists that can’t be solved by an optimizer. The other camp would not touch the optimizer with a 10 foot pole and sees it as little more than a gimmick. Where’s the truth about supply chain optimization? As usual is lies somewhere in between.

In fairness to both camps, the examples to support each case are numerous. There are plenty of successfully optimization projects and more than a few abject failures out there. With examples on each side, it’s understandable to be nervous about jumping into an optimizer project. While the potential for savings is massive, the risk of implementing a failure is a bit intimidating. So why even bother? Or better yet, what scenarios might be a better case for an optimizer project due to bigger returns and manageable scope? Identifying why you might need an optimizer and setting expectations accordingly will focus the entire organization around a tangible deliverable, rather than promising that an optimizer can save the world, one run at a time.

Here’s why you need an optimizer.

1. Manage Capacity Constraints

The optimizer does a great job managing restrictions on available resources. That means that for a business with either limited capacity, or seasonal demand, the SNP optimizer can help make the most out of a limited number of hours. The solution of the optimizer is cost based, meaning it will produce a solution with the least possible cost. That does not mean a perfect solution, instead it means the best possible solution with the given scenario. So if there just isn’t enough time to make the demand at hand, the resulting solution will make the best of a bad situation, depending on the inputs from the supply chain modelers.

Based on the costs to carry inventory, and the cost to short demand, an optimizer can help to objectively manage the delicate balance between the two. There are definitely nuances to this balance. Differentials based on demand types, demand priorities and cost of shortfalls and safety stock cuts can all be managed to derive the preferred balance. When done properly, it becomes easier to get ahead of your inventory strategy and manage capacity constraints ahead of time. Different scenarios can produce different results, adding overtime (at additional cost) adding capacity, enabling co-pack partners or different sourcing options can all be modeled effectively with an optimizer.

2. Leverage Multiple Sourcing Options

When multiple sourcing options exist for the same products, the problem is not usually where to put the cheapest product, but how to minimize the impact of making the most expensive product. This is really where the SNP optimizer can earn its keep. When done by mortal humans, this type of optimization problem results in a workable solution, not an optimal solution. It is nearly impossible and definitely not efficient for people to do this type of analysis in an objective manner.   The multiple sourcing combinations and the impact of one decision on the next is akin to the infamous “butterfly effect” resulting in unknown ramifications down the line.

People also have their own biases which would limit the sourcing options available. “Product X is always made here” limits the options and could possibly prevent the discovery of some new sourcing options that could unlock some overall cost savings for the organization as a whole. The optimizer solves the entire problem at once, which looks at all possible combinations to derive the least overall cost to the entire problem. When evaluating these results, it’s important to see it as a whole.   Costs at one location may end up rising more than before the optimizer was used. However, when viewed as a whole, the increased cost to that location results in an overall benefit for the entire supply chain. It is critical for the organization to prepare for these types of scenarios prior to adopting the optimizer. After all, if the supply chain wants to run as it always has, then the results will be the same and the optimizer can’t change that.

3. Maximize Manufacturing Efficiency

Even within a production location, the optimizer can maximize efficiency of a production plan. When using the SNP optimizer purely for production optimization (disabling external procurement) it can be used to maximize efficiency of a producing location as a whole. Locations that can make similar products on multiple lines benefit from an optimizer. Products with long changeover times, due to allergen cleanup, or die change also benefit from planning via an optimizer. When changeover costs are introduced via the setup matrix, the optimizer can balance the cost of the changeover against the cost of carrying the additional inventory, resulting in the optimal prebuild of inventory to offset the expense of a complicated setup.

Another scenario where the optimizer comes in handy is what to do with “extra” semi-finished goods materials when they are made in large discrete batches. In this case, the finished goods demand might require 600lbs of an input. The batch can only be made in 1000lb increments due to manufacturing constraints. So what to do with the extra 400lbs? A properly configured optimizer with excessively high storage costs on the semi-finished goods will actually recommend which finished goods to produce with the extra 400lbs. Whether pulling forward demand, making inexpensive products or building more of fast movers, the optimizer will convert this semi-finished goods inventory into finished goods in the most efficient manner possible, taking into account the cost to manufacture and store the finished goods.

Final word of warning

A word of warning: an optimizer is a harsh mistress. It’s cold blooded in its quest to find the lowest possible cost. As I’ve been told, “The optimizer will sell its own grandmother for a nickel” and that’s true. It could possibly rearrange your entire supply chain to save a dollar. That’s where people and common sense come in. The optimizer is only as good as the modeling team in charge of it. This modeling team needs to review the results against the reality of the day, and manage costs to avoid the impossible, as well as remove options that are mathematically efficient, but impractical in real life. This is the key that most failed implementations forget to add: the people. The optimizer is just a tool; in the hands of the right people it is very effective; if left to its own devices, it is nothing more than an expensive gimmick.

For an in-depth look at the optimizer, watch our on-demand webinar recording: SNP Optimizer: How to Approach, Manage and Maximize the Benefits, presented by myself, Claudio Gonzalez and Sanjay Jelaji.

You can also check out Sanjay Jelaji’s recent blog: Making Sense of an Optimized Supply Plan

Or you can check out the SNP Optimizer webcast we recently did with ASUG.  Click here for the slides and replay.  (Note that you’ll need an ASUG login to access the page.)