Skip to Content

Hungry once again. In my last blog we discussed pastry. Today I feel more like pizza – and to order in instead of making pizza myself. I browse to my favorite pizza delivery service’s homepage – and I am faced with a fundamental decision:

  • Shall I start to build my very own pizza (“Watch the pizza of your wildest dreams come to life.”)?
  • Or should I pick a specialty pizza (and assume the chef is a better pizza composer than I am)?



Picture courtesy of based on the pizza ordering app from domino’s


Ah, this is a difficult choice. Thinking of it, I would actually like to choose a predefined specialty (the “phantasia chef”), get rid of the onions and add some extra olives. Whoa! This looks delicious.


How about ordering a steel strip for dessert?

Actually many of the choices and the modeling tools we used without even realizing to order our pizza are core concepts of the master data model in mill products and mining. In my previous blog we discussed the use of batches, a lean modeling option below the material master, that allows to model individual pieces or similar items in inventory and describe them by attributes.


In this blog ´we will take a first look at variant configuration, and how and why we typically use it in the mill and mining industry. And yes, I intentially include the mining industry. In all cases that I am going to describe here variant configuration is the means to describe customer demand (beyond a mere product code with additional attributes).


In a dimensional product – like a steel strip –  you will find length, width, thickness, but also further attributes describing surface treatment, grade – depending on the specific industry vertical and product.


But do you really need variant configuration for this – couldn’t you just model  by a bigger number of “normal” product codes?

I have a clear opinion on this one which holds for most use cases in our industry. But I will also point out a few examples where I personally go for the simple option of an SKU.


I suggest we take a brief tour across reality-proven use cases for variant configuration – starting with a very simple product model – in mining.


Mining – what is configurable about concentrate?


As described in SAP Best Practice for Mining contract to invoice scenarios for coal or concentrate – we recommend to capture the customer’s specification of the required grade in the variant configuration of the sales contract. There are no object dependencies, configuration rules in


this model. Just a few attributes to capture the allowed ranges for moisture, the requested content
of metal in the ore, and the tolerated level of impurities and unwanted other elements like e.g. arsenic.

Later in the process this is used to match the demand attributes of the sales order config against the various stock piles and their supply attributes.



Furniture – zillions of options for high end consumers

1)   Variant configuration in the furniture industry is much a different animal. On the lower end of complexity, it can be used to describe a fairly simple product like a chair. With armrest, or without? Which colour of upholstery? Which colour of legs?

You will already find some rules controlling the allowed combinations. More interestingly, this reaches into the manufacturing definition of the chair by determining the bill of materials, and the routing based on the attributes and some further rules.
/wp-content/uploads/2013/02/273013_l_srgb_s_gl_180260.jpgOn the upper end of complexity (in furniture) take a high end custom made kitchen. Here the variant model  is multi-level (a kitchen consists of cabinets contains drawers), and needs to talk to CAD or other customer facing 3D-visualization. Many of the rules are modeled in variant config, some geometrical constraints in CAD.
If you need to model this, get yourself an expert and become a member of the Configuration Workgroup.
Independent of the software used, this is highly complex modeling.


The rules and the model itself are very much “discrete” and not so much “mill” – as you would see e.g. in metals or paper.




Metals bars, tubes or strips – MTO or not MTO?

In the metals industry I see a strong move to make-to-order configuration, at least in the high cost regions of the world. A customer can order any allowed combination of dimensions within a certain range. Some producers allow only steps of 5 or 10 mm, others give you the full freedom. In any case you get a virtually endless number of mathematically possible combinations.

Picture courtesy of Salzgitter and
World Steel Assoc.

A tube has a fairly simple geometry. Inner and outer diameter, length – that’s it.

As long as you sell only a few standard lengths, this can still be managed without variant configuration. But as soon as a customer can order any length, this gets quickly unmanageable with plain SKUs.


My clear recommendation is to use variant configuration also for this. I know a few examples where customers still sold mainly predefined products with predefined dimensions. If the number is relatively small, variant config is sometimes considered an unwanted overhead. In many cases, over time the product range expanded, the model changed from make-to-stock to make to order over time – resulting in a dramatical increase of the number of SKUs.


Better start simple and small with a very reduced variant config model, but the ability to grow with the business and your customers requirements.


Cable is a very similar use case. Its complexity depends on how you manage length. Do you sell only standard length, or any individual length? Do you manage individual lengths also in inventory?
Suppliers into the hightech or automotive industry sometimes add connectors to their cables. One connector – a piece of cable – another connector. Just this resulted into 14 million possible combinations in one specific customer case (see customer presentation at the SAP Configuration Workgroup).


Flat steel – high variations, complex requirement specification and how to track the indidual piece


What makes this specifically tricky are 3 things:

  • The production process itself.
    The steel making, casting and rolling creates always slight (or large) deviations – so the actual attributes describing the production result differs always from the customer’s target attributes in the sales order.
  • The individual piece. Throughout rolling and slitting each coil becomes an individual – with its own attributes. A sales order for 60 coils (all with the same sales order config.) will be produced into 60 very individual items with own item characteristics like dimensions, grade quality, defects etc. Some may have common manufacturing process parameters, and there is also a traceabilty flavor to this: each has a neighbor, predecessor or successor.
  • The complex requirement specification in the sales order. Typical in steel is also the combination of specifying a customer’s requirement by an standard industry steel specifications, which is internally matched against one or more inhouse grades.


Aspects like the manufacturing and supply chain planning rules need specific attention. Where (on which machines, or route through production) can you actually manufacture a certain customer specification? How do you need to control quality management based on the customer specification? You will find different solution implemented to model this at steel companies. In most cases a combination of variant configuration and batch management are important building block for a solution.


Paper, textiles and all the rest


Paper rolls or sheets are similarly 2-dimensional like metal coils and strips – at least from a modeler’s perspective. In textiles, yarn is defined mainly by length. Fabric is more tricky. For ERP it may be just another 2-dimensional product with a pattern, a dye or colour. In the eyes of an textile engineer there is much more to this, especially in manufacturing and planning.


Back to pizza

We started with 3 options to order a pizza:

  • Shall I start to build my very own pizza?
    This is our classical normal variant configuration starting from a KMAT (a configurable material). You are free to choose from all options allowed in the model. But you also need to specify most of them. Could be a little cumbersome.
  • Or should I pick a specialty pizza?
    This is selecting a material variant. All options are predefined. Accelerates the ordering of your pizza dramatically. But your stuck with what the menu offers.
  • Or should I further customize a predefined specialty?
    This is what we call in mill products a configurable variant. Quick ordering process, plus lots of flexiblity.
    for European eyes, you may seem to be a little bit picky to the waitor if you redefine everything.


Configurable variants?

In most other industries that use variant configuration, you have configurable materials (KMAT) that offer you a nearly infinite number of possibilities. And you have a material variant that is exactly one possibility out of these described with an own material number. We call this a fully configured variant, or sometimes a make-to-stock variant.

In mill products we often use another kind of material variant. We call those configurable variant, or sometimes make-to-order variant.
They “belong” to the configuration model of a KMAT as “ordinary” variants, but they are still configurable. Think of them as prefilled templates, where most characteristic values have already been defined, but only a few are still left to enter e.g. the dimensions. As such, configurable variants accelerate sales order entry, and make the KMAT’s model for tangible. I will take about further benefits like costing, pricing, inventory visibility another time.


Configurable variants are part of SAP standard in ERP. They do not even require the industry extension for mill products.

Want to know more? I will discuss usage of the various type of variants in my next blog. See you then.


My conclusion?

I like variant configuration. From my experience, it allows a much more manageable master data model in all of the above use cases. And in some cases, it may be fairly impossible to model without variant configuration. Does this also reflect your experiences? (And did you ever attend a CWG conference?)

To report this post you need to login first.


You must be Logged on to comment or reply to a post.

  1. Fred Verheul

    Nice one Stefan!

    As a developer I’m totally not into this Mill-industry-specific thing, but this blog was easy to read, ánd I learned something.

    And yes, I’d prefer the nonexistent (at least in the Netherlands and to my knowledge) configurable variant pizza 🙂 .

    Cheers, Fred

  2. David Purslow

    Hi Stefan,

    excellent article with some great examples to explain away the intricacies within the whole VC subject. My site uses Material Variants in the traditional way and I have suggested trying the different examples you have spoken of (we make a wide variety of instruments and sensors with many manufacturing models) but there is only me 🙁 that gets excited about the different ways to set up – a lot of the engineers and managers think the VC subject taboo as they perceive it complex and “black magic” – so thanks for removing the darkness.


    1. Stefan Weisenberger Post author

      Hi Dave,

      thanks for your nice comment. Although the VC topic is itself a bit complex – independent of software tools used – there are a number of best practices. If you would like to discuss this in a group of peers – why not attend the 2015 Fall Conference of the Configuration Workgroup and discuss your ideas face to face with the experts.

      Thanks and regards, Stefan


Leave a Reply