IT in Manufacturing

Enterprise manufacturing intelligence (EMI) in process industries

December 2012 IT in Manufacturing

Relationship between business intelligence (BI) and EMI

Enterprise manufacturing intelligence (EMI) or operational intelligence (OI) as a concept has been slow in being adopted within manufacturing. Most plant managers and chief operating officers want the information contained in EMI solutions, but they cannot motivate EMI against traditional business intelligence (BI) technologies applied at enterprise level or they think it is the same thing.

EMI and BI have different purposes; and they are aimed at different audiences. Manufacturing-specific operations reporting and intelligence is different in content, context and data frequency than the data in BI. BI tools typically focus on financial and KPI-level analysis and reporting, where EMI solutions look at manufacturing-specific KPIs as well as the key process influencing factors (KPIFs) related to those KPIs.

Data in a BI solution is typically at the same low frequency as that of the ERP system such as daily values. For a plant manager that wants to know what is happening on a shift or hourly basis, BI will thus be inadequate. BI tools are typically not designed and implemented to take into account the real-time nature of manufacturing operations, and the resultant very large data rates. As such, BI is not able to handle the high frequency of data receipt and the required fast response-times of reporting/visualisation required by manufacturing operations.

Executives use BI as strategic analysis and decision-making tools for the company. From their BI systems, they can see the profitability of individual plants and sites and, as such, can make the decision to close down a plant or to change the manufacturing strategy. They typically work on confirmed and validated numbers and results as they want to ensure they have accurate data when they make the decision. These validation or auditing steps often add considerable time between the actual event and the time the data end up in the BI solution.

Site-level production personnel however cannot wait for the niceties of auditing and validation before they take action (see the time-value of information section earlier in this document). If a report or an EMI dashboard indicates that something is wrong, it is their responsibility to investigate and take corrective action. If a feed-rate is lower than planned, the production manager is not going to wait for the confirmed result in the BI system tomorrow before he takes corrective steps. No, he is going to investigate or have someone investigate for him. If it turns out to be a false alarm, then he is glad as it is a crisis averted. If something is wrong, he takes corrective action, or at least knows and expects the bad results from the BI system tomorrow.

EMI systems thus have a two-fold purpose:

1. To provide early warning in real-time for potential problems in order to make decisions or take action.

2. To provide ‘slice and dice’ on historical data for process improvement.

EMI has data available at the granularity and frequency delivered by the individual applications. This can be from seconds to days, depending on the specific operations requirement. The data is also available per individual piece of equipment, line or processing unit and can also be rolled up into hours, shifts, days or weeks for any of these. The granularity of EMI systems is closer to real-time and they are often used as real-time dashboards for operations executives.

BI may be able to provide the historical ‘slice and dice’ data, but typically not at the level of granularity required by operations managers. BI will not be able to provide the real-time early warning required by the plant. Both of these are thus needed to support manufacturing companies adequately.

Key process influencing factors (KPIFs)

KPIFs are not typically reported at enterprise level. They are not considered as KPIs as individually they have no financial impact on the business. KPIFs may however form part of an EMI implementation, as these factors influence the performance of financial KPIs.

An example of a KPIF is for instance temperature in an endothermic reaction that influences the yield (the KPI) of a specific process. If the temperature is too low, the reaction will be impeded and equilibrium will be reached before the reaction can be completed, leading to lower yield and raw-material losses. Process yield can be related to financial results directly (therefore being a KPI) but temperature cannot (as such not a KPI). The temperature of the reaction however can have a major influence on process yield. It also does not mean that if the temperature is controlled in the right range at the expense of all other variables that the yield will be better, as there may be other process influencing factors that can also affect the yield (such as contact time, pressure, agitation and shear).

KPIFs are typically stored at more granular frequency for use by process engineers and other production personnel. KPIFs are best evaluated when the values are being viewed as trends and not as a point in time. EMI solutions are thus well suited to display KPIF values as they make provision for more granular data within a context of other KPIF values.

Discreet vs continuous and process industries

EMI tools are used in the discreet industry to visualise and analyse manufacturing data in such a way as to measure performance against a schedule or plan and to determine the resources consumed to produce a number of units. EMI in this case can be used to map manufacturing indicators directly to financial indicators, using for instance resource consumption accounting.

In the continuous process industries, EMI solutions are used very differently as the final output from a process is less as a result of planning the raw-material feed into the process, and more a case of trying to abstract as much value as possible from the inherent value within the feed material.

As an example: In a discreet process, 1000 units of component A is combined with 2000 units of component B to produce 1000 units of product C. If we double units A and B we will get double the product out. The units can be counted, touched and seen. There may well be some losses due to quality or process issues, but typically these are very apparent and easy to predict.

In a continuous process however, things are less defined. 100 kilolitres of material A is received at 25% concentration. This material is reacted with 250 kilograms of material B at 99,5% and 50 kilolitres of material C at 60% to produce 450 kilograms of product D at a theoretical yield of 96,43% over a 24 hour period. In this process, pressure, temperature, feed-rate and residence time play a major role, and depending on these factors the actual yield can vary between 83% and 95%. This process is thus a lot less predictable than a discreet process and different tools and management philosophies are required to get the best from the process.

In the discreet process above, the EMI solution will show for instance the number of units being produced per hour, the number of rejects, the energy being consumed and the predicted time when the order will be completed. These indicators can all be directly tied back to financial indicators and costs.

In the continuous process, the EMI will have to display the specific feed-rate of the different materials, the temperature of the reactor, the pressure in the reactor, the pH of the solution and whether these are all balanced within the ultimate process state. None of these indicators have any relation to financial indicators or costs, but they do influence the ultimate yield of the process dramatically. In this case, the EMI solution will also have to display the KPIFs in addition to the KPIs.


Looking at the above, it is clear that EMI is not the same thing as BI and their actual use and benefits are not the same as they are aimed at a different user community. It also shows that an EMI within a discrete industry or a process industry will not necessarily show the same type of indicators, nor should it. There will always be overlaps as with most solutions and architectures, but as with any similar situation it is up to the users to determine and decide what they want to show and in which application.

For more information contact Gerhard Greeff, Bytes Systems Integration, +27 (0)82 654 0290,,


Share this article:
Share via emailShare via LinkedInPrint this page

Further reading:

Navigate the fourth industrial revolution with PricewaterhouseCoopers
November 2019 , IT in Manufacturing
Using the building blocks of 4IR to transform business processes into manufacturing advantages requires a holistic approach.

Artificial intelligence in manufacturing – a practical and simplified view
November 2019, Altron Bytes Systems Integration , IT in Manufacturing
Looking at and interpreting data generated during the manufacturing process to find ways to reduce waste, improve quality and increase yield is not new. However, the increased use of digital technologies ...

Intelligent data glasses support production
November 2019, Siemens Digital Industries , IT in Manufacturing
Augmented Reality (AR) has arrived in the workplace: for the last three years, a consortium of six companies and institutions chaired by Siemens has been researching the use of augmented reality (AR) ...

Siemens launches report on Dawn of Digitalisation and its Impact on Africa
October 2019, Siemens Digital Industries , IT in Manufacturing
Africa needs a digitalisation strategy to compete in the global economy.

Integrating IT with Operations Technology - a CIO perspective
October 2019, Absolute Perspectives , IT in Manufacturing
Digitisation projects, spurred on by Industry 4.0 and technological advances, have opened up many possibilities to create new value.

Real-time data increases metal recovery at Peñasquito mine
November 2019 , IT in Manufacturing
At a recent OSIsoft User Conference in California, ARC Advisory Group had an opportunity to learn about a related project at Newmont Goldcorp’s flagship Peñasquito gold, silver, zinc, and lead mine in ...

The power of data
October 2019, Throughput Technologies , IT in Manufacturing
Gain access to important information with ProSoft's new OPC UA-based gateway and data logger.

Securing and managing devices the modern way
October 2019 , IT in Manufacturing
Organisations of all types and sizes are bringing vast numbers of devices into their IT infrastructure and networks. Moreover, these devices are not only smartphones, wearables and tablets because the ...

Reliable gas production
September 2019, Siemens Digital Industries , IT in Manufacturing
Energy supplier uses Siemens technology for optimum information exchange.

Automotive supplier relies on Siemens PLM solution
September 2019, Siemens Digital Industries , IT in Manufacturing
Küster Automotive is an internationally active family-run company and a part of Küster Holding. A full-service supplier, Küster manufactures components for the automotive industry. For its production ...