Aug 18 2011

Print this Post

“Myth or Reality: Is Manufacturing Intelligence Just a Simplified Version of Business Intelligence?”

This is quite a common question – what is the difference between Manufacturing and Business Intelligence (BI) applications? Since many manufacturers often have a BI solution already implemented, the first thought is that BI can simply handle their manufacturing data … so do I need to add a manufacturing component?

This is a similar discussion on whether Enterprise Resource Planning (ERP) systems can be used to run manufacturing operations, instead of a Manufacturing Execution System (MES) or Manufacturing Operations Management (MOM) solution. Some have tried, many have failed.

Why do I suggest this parallel? Because there is a common theme of trying to use a corporate system to address a local, shop floor need. In fact, it looks like we have another “Myth” that just might need to be “busted.”

To start, let’s define the challenge: Can I address all Manufacturing Intelligence requirements with Business Intelligence solution?

The situation: Let’s say we are a manufacturer with several sites around the globe, and we need:

  • Real time visibility into manufacturing operations data, including logistics, production, quality and resources
  • KPIs that can be calculated and visualized, starting from individual resources aggregating up to the factory level
  • Analysis performed on correlations between various production tasks from different functions, from different perspectives
  • Propagation of the information between systems and people (including shop floor operators)
  • Real-time answers on ad-hoc questions raised on the shop floor dictate 24/7 availability
  • Assurance that all necessary data is readily available to support regulatory compliance

The test: All of the above requirements are typical and can be addressed by a Manufacturing Intelligence system, so there is no need to test. Let’s see how Business Intelligence can address them by looking at some specific aspects.

Real time integration with shop floor data sources (machines and systems)
BI is typically implemented at the corporate office to identify, extract and analyze business data related to sales, revenues and expenses. Even though BI offers a wide range of connectors to take data from different sources, rarely can those systems acquire data in real time, especially when operating remotely.

Issue #1: Availability/reliability of BI is not adequate for real time MI data

Data model includes manufacturing operations
Manufacturing data from the shop floor (MES/MOM) are typically much higher in volume and lower in granularity than financial data in an ERP; BI systems operate similar to ERP, as they are typically batch oriented with an analysis range that spans from days to years.

Issue #2: Data volume is too great for BI applications to adequately analyze and manage

High availability system accessible from the shop floor and corporate
As BI is typically installed at the corporate headquarters, but this architecture is not well suited for the shop floor. One issue is security, when IT separates shop floor networks from corporate systems. Even when users can login into the BI application from the shop floor, too many layers and systems are required to pass data to and from the central BI application, resulting in processing delays, reducing the effectiveness of analyses performed.

Issue #3: IT policies and architectures create boundaries that prevent easy access to contextualized shop floor information

Ability to present data in context of end user activities
Many BI tools are very mature in data presentation and flexible in adaptation, but this doesn’t help those working on the shop floors that need to contextualize their analysis of data with the actions they are performing. Information needs to be easily accessible from within a familiar application environment, such as a MOM or MES system.

Issue #4: People do not want to switch between applications to perform ad-hoc queries

Conclusion: Similar to the debate between MES and ERP, there are similar discussions between BI and MI systems. But, as the above examples demonstrate, when you look under the outer “bodywork” of BI, it quickly becomes clear that it isn’t really a substitute for a true Manufacturing Intelligence system. Of course, one can claim that customizations are always possible. This is the same argument we have heard about ERPs. Is it worth it to try and change a Hindustan Ambassador (car on left) into a “wanna be” Peugeot 206 (car on right), when in the end, you still have an Ambassador?

If you haven’t yet seen this video advert, I recommend you do so.

Myth busted. In general, there are too many gaps for a Business Intelligence application to efficiently cover all Manufacturing Intelligence requirements.

Permanent link to this article: http://www.apriso.com/blog/2011/08/%e2%80%9cmyth-or-reality-is-manufacturing-intelligence-just-a-simplified-version-of-business-intelligence/


1 ping

  1. Xaria

    Walinkg in the presence of giants here. Cool thinking all around!

  2. Tadeusz Dyduch

    Thanks for the comment. I’m glad to have this excuse to dig down into reasons and scenarios I have in mind.
    First of all I agree that single MES/MOM system implemented across the plant provides a lot of Manufacturing Intelligence functionality – data aggregation, contextualization, propagation and visualization.
    It can also greatly simplify the use of any 3rd party intelligence application and in some cases even corporate BI could be enough to cover remaining scope (in some industries).
    However the reality is that still many manufacturers use multiple isolated systems and even a broad MOM deployment takes time.
    In that case, MI needs to plays a role of a real-time communications hub at the factory layer.
    Corporate BI is too much of a centralized beast to provide this, especially in heterogeneous environments or in factories distributed around the globe.
    Quality and diversity of data + a higher need for continuous changes are common reasons why IT organizations avoid moving factory level MI functionality into a central BI system.

    Whats more, MI provides some additional values I have not mentioned before – well designed, can be used for shop floor data archiving.
    It can be used for short and long term analysis/reporting that is consistent and easy to do after many years since operations were executed.
    Still, BI can provide some other values at the centralized level sourcing sanitized and well formatted data from MI. Overall, this gives flexibility that plant level people need and visibility that is demanded by HQ.

    As for the last point (ability to present data in context of end user activities), I have two MI users in mind.
    First would be the process engineer that performs more off-line continuous improvement or root cause analysis. He can use BI under the assumption the granularity of the data is enough.
    Second would be line supervisors or even operators that need information more real-time and in the context of their current activities to be able to quickly react to some shop floor incidents.
    The more easy and contextualized are the information the more self-efficient are people at the shop floor and decisions are taken faster.
    Again, some MES/MOM solutions can provide much information, but not with the interactivity provided by MI applications.
    The ideal situation would be to have one MOM solution providing also the MI functionality at the plant level and exposing well structured data for corporate BI.
    I discussed this a few months ago if you wish to read: http://www.apriso.com/blog/2011/04/the-many-benefits-of-data-unification/
    Again, thanks for sharing your comment. I would be happy to see if those examples comply with your experience.

  3. Delcuvellerie

    Very interesting topic that we are precisely investigating inside our company. I would challenge your assumptions:
    – 24/7 availability: we don’t see this as a pre-requisite as shop floor reporting is already available in our MES applications and sufficient for immediate execution. MI (Manufacturing Intelligence) analytics are considered for off line ad-hoc analysis as part as continuous improvement initiatives.
    – Real time integration with shop floor data sources: again our MES applications are doing that job of collecting shop floor data so all manufacturing execution data are available in MES SQL databases. So thay can be easily accessed from any BI tool. Most if not all BI tools can run continuously with short term data (whatever if it is with sales, stocks transactions data or manufacturing data).
    – Ability to present data in context of end user activities: ad-hoc analysis is usually not done by operators but production or process engineers, so assuming the BI/MI application is properly designed with the right combination of master and transaction data, there is little need to have an integrated GUI with the MES.
    Nevertheless I would agree with the other assumptions. Let us know what you, as well as your other industry users think about this.

  1. Making Sense of Enterprise Manufacturing Intelligence | Manufacturing Transformation Blog

    […] First of all, what works for the enterprise might not work well on the plant floor — these environments use completely different metrics and processes. So, the rule of thumb #1 is: don’t expect a business intelligence system — even with a manufacturing component — to be able to effectively measure the variables on the plant floor. This BI vs. MI debate has been explored in greater depth in this earlier post. […]

Leave a Reply to Making Sense of Enterprise Manufacturing Intelligence | Manufacturing Transformation Blog Cancel reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>

Time limit is exhausted. Please reload CAPTCHA.