What's Hot

Nassim Nicholas Taleb's blog, an inspiring read | Incerto

Thursday, December 22, 2011

Risk Balanced Selection of Manufacturing Projects

Question: I have a situation in manufacturing operations where we have to take a judgmental decision on our equipment’s re-furbishing based on visual inspection. It is a very crucial area of the business since it is part of the complex and risky Klinkerization process.

In the short of it, we have to decide whether to replace brick lining which is critical for the survival of the equipment before the next shutdown and that is a year later.

There are several ways I would go about dimensioning risk on this type of potential project and these have been introduced in the presentation that is attached to this post.

These type of decisions are often difficult to make because of the relationships between costs, profit and risk. Additionally there will be alternate stakeholders involved in the decision making process and we can nearly guarantee that each stakeholder will be focusing on one aspect of the program over another. Increase the stakeholder pool and the motives generally go up proportionately.

In the domain of operational risk we have several tools that help us form coherent decisions or perhaps risk based decisions if you may. In effect we need to ensure that the business enters into a choice to change an environment that is based decisively on known information and that it accepts the potential and probable fiscal impacts from engaging in such a change decision.

There are five different types of  analysis we may entertain which paint different pictures of the potential outcome space and they are also contextually balanced. Being contextually balanced furnishes us with different views on why we choose what we do.

  5 Activities that should be considered

Is our motivation based on a cost saving or risk aversion or profit and, how much does each choice potentially cost us in risk?

In the presentation attached we introduce these five quantification concepts. However, this document should be seen as an introduction or overview of these concepts, and these methods will need to be evolved as singular exercises in their own right separately.

The presentation can be found at this link

In the context of this problem
The choice on which model to use will be based in some respects on how much time you have before making the final decision which is : Change the brick lining or not. 

Questions to ask yourself (there are many) but some include:

Can the change request be carried out in phases, maybe, maybe not. One choice might actually involve building a second furnace alongside the main facility but again that comes down to costing and the real option price to entertain such a double up investment. I can pass a model to you on this.

If you have more time (ie a month or two), I would entertain a Bayesian method or a Real Options approach, even both for forming this business decision.

If time is short, then a scenario analysis method might be the way to go. In some respects it might be good to start with a scenario analysis approach anyway and then work back from that framework towards a Bayesian decision tree. In this way you have something to work with up front and the scenario analysis approach will frame potential network nodes you might want to consider modelling in the Bayesian Decision Tree.

Always remember with the scenario analysis approach that there is a null scenario that has to be investigated. The NULL scenario is NOT engaging in the change-request, the decision tree starts at that point.


  1. Thank you for posting the great information. The information mentioned in the blog are some of the best Manufactur suggestion available.

  2. Thank you for excelent Information,