Why is it that after investing hundreds of thousands of dollars in a shiny new enterprise resource planning (ERP) system, you still find yourself looking at spread sheets when trying to figure out the best plans for production, purchasing, inventory and distribution—decisions that have a huge impact on your bottom line?
The truth is that most ERP systems simply were not built to cope with the challenges of a food company, which are quite different from other companies, such as a car manufacturer. Most ERPs struggle with things like managing shelf life, finding the ideal batch sizes, maximising yield or dealing with reverse recipes and/or reverse material requirement planning (MRP).
The tools most companies use for planning today, such as MRP, older generations of advanced planning and scheduling (APS), or even the most common approach, manual planning in spread sheets, all have one thing in common: they do not offer the planner any kind of ‘intelligent’ decision support, but are based on simple, pre-defined rules (or ‘heuristics’ as they are sometimes referred to).
The planning is sequential (one constraint at the time) and separate (one plan for purchase, another for production), and the rules that you can pre-define are limited. As a consequence, you end up with un-realistic plans far from the optimum and your company is not as efficient, productive or profitable as it should be.
Optimisation is a word that is being used quite loosely and in many different contexts, but almost never in the strict sense applied here, namely the use of mathematical algorithms (linear programming) to identify the plans that will minimise the total supply chain costs.
Simply put, optimisation will not only find a realistic plan, but the very best possible plan. This is a huge difference from older generations of planning tools and something that can help increasing your business’ profit, productivity and competitiveness.
Mathematical optimisation has been around for a while, but it is not until recently that it has become readily available as a support for your operational planning. This is explained by the introduction of the 64-bit version of Windows, coupled with the latest developments of optimisation algorithms and computer processing power.
Too Simple For Optimisation?
Ollie Jones, US
You may think that your business is not complex enough to gain any substantial benefits by using mathematical optimisation. Consider then the following simple example:
Imagine Burt the Baker, on the night between Wednesday and Thursday, standing in his bakery trying to figure out what to bake in order to make as much money as possible in the coming day.
To keep this example simple, Burt has no capacity or other constraints to consider, only the availability of ingredients. To keep it even simpler, we only have two products and need not worry about the day after tomorrow.
We will make it even easier by simplifying the recipes for these two products: No need for complex optimisation for such a simple case, right?
So, how many breads and cakes should Burt make in order to maximise the total contribution? He certainly cannot make 45 breads and 25 cakes, because he simply does not have enough flour and sugar.
Burt scratches his head, looks at his spread sheets and since cakes have the highest contribution margin for every sold unit, he decides to make as many cakes as possible and use the remaining ingredients for bread. This earns him a total contribution of $800.
Burt makes similar decisions day after day, year after year, and since at the end of the day he makes a decent profit, he does not see any need to question what he is doing.
In this particular town, there are two other bakeries. One of his competitors, Conrad the Cakemaker, has tried to improve his business, by first implementing MRP, and when that did not offer enough improvements in the planning area, he then invested in an APS as well.
With the same basic conditions as Burt’s, Conrad’s MRP proposes to make 45 breads and 25 cakes, but then his APS system automatically detects the material shortage, and since Conrad has set up a rule that says bread has the highest priority, the system re-schedules the planned manufacturing order for cakes to Friday, and informs Conrad about the missed demand.
Conrad does not think this is a good solution, and therefore, decides to manually re-schedule. He makes the same assumptions as Burt, and makes as many cakes as possible, and then he uses the remaining ingredients for bread.
The third competitor in town, Optibake, has recently started using an optimisation tool for improved decision support. With the same conditions and limited availability of ingredients as his competitors, Optibake’s planning tool transfers the planning challenge into a mathematical expression:
An optimisation model approaches the planning problem in a completely different way. There is an ‘objective function’, which describes what we are trying to achieve. The objective function in this simple example only contains profitability.
In addition to the objective function, we also define all the constraints we need to consider to get a realistic plan. Once this is done, the optimisation engine will now evaluate all possible solutions until the best one is found (for most companies, the best solution is the one that gives them the highest profit).
When all is said and done, even in such an incredibly simple example as this, through the use of optimisation, Optibake made a 25 percent higher profit than its competitors ($1,000 compared to $800).
Optimising Your Food Business
Spengler, Brighton, UK
As you saw from this simple example, mathematical optimisation offers true decision support for your planners that not only simplifies their life, but actually helps you make as much money as possible. And if that is evident from such a simple example, imagine what optimisation can do for your business, especially with all the complex planning challenges your planners are facing on a daily basis.
The beauty of optimisation is that it will tackle all the constraints that you decide to include at the same time and in parallel, in order to find the plans for purchasing, production, inventory and distribution that will minimise the total supply chain costs.
In addition to the classical planning challenges of making sure that we have sufficient line and operator capacity, as well as ingredients and material, optimisation can be used to solve some of the more difficult challenges that you may face in your food business, such as:
Shelf Life: For products with shelf life constraints, the optimisation engine considers the expiry of products as a cost, and therefore does its utmost to avoid it. An important step in this process is to warn planners at an early stage.
Batch Size: The optimisation engine is able to determine the ideal batch sizes for each period, by balancing the cost for production changeovers, with the cost for inventory and missed deliveries.
Seasonal Variations: The optimisation engine is able to find the best way to handle seasonal variations, by balancing the cost for building stock with the cost of increasing the capacity, either by adding extra shifts or by using alternate supply sources.
Yield: When the ingredients can be used in alternate ways, the optimisation engine will determine their best usage by comparing the supply to the demand, and the cost for these alternate usages. Optimisation will also try to minimise the start- up scrap in production.
Divergent Product Flows: For divergent product flows, where raw milk becomes fresh milk, butter, cream, or where an animal is split into different end products in a variety of different ways, the challenge is to find the perfect balance between demand and available supply that will maximise the yield, the delivery service and the overall profitability.
This is managed through reverse recipes, with different costs for alternate ways of processing the supply.
Reverse MRP: For many food businesses, the traditional MRP found in an ERP-system is close to useless. For perishable ingredients, which must be consumed within hours or days, there is no point in trying to break down the forecasted and actual demand, as in MRP, but planning rather becomes a matter of determining the best way of ‘pushing’ the available supply through the supply chain.
An optimisation engine is able to do just that, while finding the perfect balance between demand and supply, with respect to yield and cost.
Product Mix: The sequence in which products are made will often have a big impact on non-productive time in production, such as cleaning. Other times, due to allergens, certain products should not be manufactured at the same time. An optimisation engine is able to identify the ideal product mix, and may even work together with cyclic planning.
Ideal Sales Prices: An advanced application of optimisation can be used for companies who are able to determine a historical relationship between sales prices and volumes.
In this case, a price elasticity curve can be established, and then used by the optimisation engine to determine the perfect balance between supply, production capacity and sales volumes, and thus the price point for each product.
These are just a few examples of planning challenges that simply cannot be managed in a good way in an ERP, an older APS, or manually in Excel.
Keep It Simple
At this point, you may realise that optimisation probably has a lot of potential for improving your particular business, but you worry if it is very complex and costly to implement.
The good news is that optimisation is certainly no longer restricted to only the largest, most successful MNCs. There are tools available that are cost effective, extremely flexible and very easy to use. You could start out small, to pick the low hanging fruit, and then add more constraints as you move along.
Quality of data is of course, always important. However, since the optimisation can take place on a slightly higher level (such as daily or weekly), and still yield great results, it is less sensitive to data quality than a traditional APS-system that may be planning down to minutes.
There seems to be a paradigm shift taking place in the area of supply chain planning at the moment. Companies who are quick to move into the future should stand a good chance of getting ahead of their competition. When considering this, it makes a lot of sense to start optimising your plans. What do you think?