Requirement specifications are often costly and challenging to create and maintain. However, if they are focused on the product's functionality, they can be a valuable tool, unlike a rigid list.

One of the problems that is often seen with products and product development is that an approved requirement specification cannot be changed, even if it no longer makes sense later in the process, or turns out to be based on incorrect assumptions or calculations. The problem is often based on inflexible interpretations of e.g. design freeze or internal decision processes or chains of command that do not support innovation.

It can be hard to change the structures in a company culture. Instead, you can use a more pragmatic and application-oriented approach to requirement specifications, where you design the individual specifications with a certain built-in flexibility.

It is not simple, nor does it solve all the problems, but it is a way of reducing the number of potential problems in both the development and production phases. Naturally there are many considerations you can benefit from making, depending on company policies, products and manufacturing technologies. The five considerations below should thus not be seen as an exhaustive list, but rather as a good place to start.

1. Requirements in regards to material properties

When you ask a company about what kind of raw materials it uses, the standard response is to give the trade name and possibly also the supplier of the raw material. What kind of material it actually is, is quickly lost somewhere along the line, and already you have set yourself up for troubles. What is it that is important for this particular material, for this particular application?  Is it its melting point, purity, toxicological profile or something else entirely?

You need to know why you are using one particular material rather than another. What properties does this raw material have that make it usable and conversely: what properties would make the material unusable?

The materials specification must therefore reflect the requirements for the material properties, not the name of the cheapest grade available on the market, fulfilling the material requirement when drafting the requirement specifications.

2. CAS number and purity

Many companies specify in a requirement specification that they will use x % of trade name A, y % of trade name B and so on. Some companies have a backup plan, so that if the raw material with trade name A cannot be obtained, then they will use the raw material with trade name C. This approach gives rise to two problems:
  1. You are then obliged to stick with one specific supplier and its prices and terms of delivery.
  2. Should any supply problems arise, or if the supplier changes its raw material, there is nothing you can do. You can minimise this problem by validating alternative raw materials, but it does not remove the cause of the problem.

The material specification of the raw material can often be boiled down to a CAS number and a purity. If you are talking about raw materials where there can be big differences between the same material, then additional requirements should be attached to the material. This is seen for example in polymers, where there can be substantial differences in molecule weight, degrees of substitution, additives and so on. This information can be added as needed, but you should start with the CAS number and purity.

Some raw materials may only be available from one supplier. This applies for example to certain unique polymers or compounds. Here too, you should start with the CAS number and purity. New suppliers may turn up and you run the risk of the manufacturer of a particular raw material making changes to the product (and forgetting to inform the customer of this). This means that companies must be acute in terms of material composition and purity.

3. Implicit and explicit material properties

Very few companies give the difference between implicit and explicit material properties any considerations. They are well aware that the same raw material from two different suppliers will not necessarily behave in exactly the same way during production, which is one of the reasons why it is a good idea to validate several raw materials for manufacture.  This is generally not associated with implicit material properties however.

The explicit material properties are the ones listed on the datasheet and the analysis certificate. While some of this information is often good to have, much of it is not worth the value which the companies ascribe to it. Two materials may well have the same tensile strength under the conditions specified by the standard used for testing, whereas the difference in tensile strengths at the application temperature, means that only one of the materials is usable.

The implicit material properties are those that comes with the product. If you purchase raw material A, then you know that the content of e.g. antioxidants/preservatives, anti-caking agents and other more or less secret ingredients have certain properties. If you purchase raw material B instead then you cannot be sure that the additives will be completely the same. So, even though raw materials A and B appear to be identical on paper, they may not actually be so. For certain process conditions then the company may never notice the difference, but for others there will be a world of difference.

The difference between the implicit and the explicit material properties will likely become very clear for companies when the raw material they have used for years without problems suddenly starts acting differently, and the supplier informs them that nothing has been changed. That is until the company conducts an analysis that shows otherwise. Then it is often, according to the supplier, a change "that is of no practical consequence". In the majority of cases this will be correct, for the explicit material properties at least.

There is nothing odious or uncommon in an ongoing adjustment of the raw materials, and in terms of the properties on which the raw material is sold, the change probably is of no practical consequence. For the implicit properties however, the change can be significant and lead to major problems with the products and/or production.

4. Entry controls

Around the turn of the century it became customary for companies to reduce the entry control to a check of the analysis certificate that came with the raw material and possibly also a visual inspection of the material. A visual inspection could for example be to check that it said the same thing on the raw material's label as on the certificate.

Entry controls require equipment and qualified personnel to operate them, so it is expensive and generally time-consuming. The supplier must check its raw materials anyway as part of the quality check, so the general opinion has been that you can just as well use the analytical data from the supplier.

In a perfect world this idea is fine - in the real world: not so much!

Analyses for output controls are also expensive for the manufacturers, especially if they show that the material is not within the specifications. We have seen examples of certificates that are not consistent with the material properties. There have also been examples of foreign suppliers that do not do any testing whatsoever, but simply fabricate a certificate. Sometimes nonconformities are detected when the material starts giving problems in production. Other times it is when someone notices that it is the same three-four sets of test results that appear on the certificates, or wonder why they continue to receive raw materials from the same batch.

There is no definitive answer to how much entry control is needed. While reducing the entry control to a visual inspection is a bad idea, the costs of conducting an entry control must also equal the gains.  One part of the solution is to identify the analyses that most quickly reveal nonconformities and be very aware of the difference between tests and relevant tests.

5. Tests and relevant tests

Conducting tests on a raw material or a subject is no problem. Carrying out relevant testing is quite another matter, as it requires insight into the test methods, their limitations and what it is that the test actually shows. 

Originally (at the start of the 20th century), people would kick the tyres on cars in order to test whether the material, which the tyre was manufactured from, was so thin that this could increase the risk of puncture. At that time this was a relevant test. You can still see people walk over and kick tyres without really knowing why they do it or what they should be looking for. Even if they did, such a test is irrelevant and useless today, and has been for decades. It has become "what you do". So:

  • Drop the redundant tests. Stop carrying out tests that do not add any real value to the inspection. There are many examples of tests that are carried out simply because the equipment is available, or the classic "this is what we always do". 
  • Get the accuracy of the analysis validated before it becomes part of a test setup. It makes no sense to use an analysis that can only detect nonconformities of 20 % or more if the max. tolerance is 10 %. Conversely, nor does it make sense to use additional time and resources on measurements with five decimal place accuracy if you only need one decimal place.
  • In general, start with the quickest and cheapest tests. You can then see whether it makes sense to conduct the more expensive and time-consuming tests. There may be some strategic considerations here that mean you have to carry out the expensive and/or time-consuming tests before you can start the quicker and cheaper tests.

The objective in carrying out an entry control is to catch nonconformities before they can cause problems in the manufacturing process and/or the product, and not to generate as many test results as possible.

Is it hard?

Both yes and no.

Technically speaking, no, it is not hard. There is a technique and a mindset, and it is not especially difficult to learn. There are companies that offer advice about how you can make flexible requirement specifications or make them for you, for example FORCE Technology.

Old habits can however be hard to change. For many people this represents a whole other way of looking at raw materials, and many will maintain the opinion that an incorrect or misleading value is better than no values at all. You should never underestimate the challenge in this.

The advice here is therefore; that the requirement specification is designed with maximum flexibility built into it, and have it based on function and properties rather than trade names.  You can save yourself a lot of money and trouble by doing so.