Manufacturing intelligence - the way of the future

An increasingly competitive environment, tolerating nothing short of manufacturing excellence, is driving the requirement for plant and enterprise-wide decision-support tools - more recently known as manufacturing intelligence.

Much has been done and much money spent over the past decade to improve quality, increase efficiencies and reduce costs, and yet on the plant floor, very little has changed; in many ways, technology has virtually by-passed the plant floor.

Effectively connecting their business systems to the plant floor, focusing on the top sources of downtime in the plant, identifying and removing constraints, and improving product quality are examples of the fundamental challenges for which most manufacturing plants still require a solution.

At the heart of these issues is the need not only to collect the appropriate data, but also to collect it in a consistent, meaningful context across the plant and throughout the enterprise. Only then can data be transformed into information that will drive significant, meaningful improvements within the enterprise.

The manufacturing landscape today

Most of the spending by manufacturers over the past five years has, ironically enough, not impacted productivity in any significant way. Many are still reeling from ERP implementations that not only did not deliver the promised manufacturing analysis tools, but actually negatively impacted production.

Supply Chain Management (SCM) implementations have not delivered the value that was anticipated, and in most cases, while some payback was achieved, it was limited to the local facility. Rarely did the SCM implementations effectively extend outside to actual suppliers and customers.

To compound this, the expenditures related directly to the plant floor, including Advanced Planning and Scheduling, Execution Systems and Maintenance Management Systems, have proven unwieldy and expensive to maintain, with a significantly higher total cost of ownership (TCO) than anticipated.

Much emphasis has been placed on various initiatives (such as Six Sigma, Lean Manufacturing, Continuous Improvement and others) that were designed to drive out inefficiencies and improve quality. And while these programs have proven effective, it's clear that much of the value derived from them comes during the early stages of their activity. In other words, those areas that can be quantitatively enhanced or improved using existing tools already have been.

The next major productivity gains will be made only with the availability of the appropriate tools and extensive operational visibility - that is, with real manufacturing intelligence. Needless to say, it's ironic that Microsoft Excel is still by far the commonest and most effective tool used by plant executives to run their plants.

In a survey conducted by Forrester in 2002, the number one response to the question, "What are your biggest problems with global manufacturing?" was "Poor visibility into plant operations." In other words, despite all the money spent on technology and various software solutions, the biggest concern among plant managers" today is "not having the right information at my fingertips to make the best possible decisions on how to run my plant".

The fundamentals of process automation

Most attempts to address issues of productivity and efficiency enhancements on the plant floor have suffered from a common drawback - their relevance and effectiveness is directly proportionate to the availability, timeliness and accuracy of data from the plant floor.

ERP

Most ERP systems depend almost solely on manually entered data for their connection to the plant floor. The most obvious concern is the time delay between the actual occurrence of a given event and the entry of that event into the ERP system. Just as significant are the accuracy concerns that stem from the probability of personal bias and data-entry errors through the many levels of aggregation at the machine, line, and department levels prior to it's entry into the ERP system.

Given the sheer volume of data required from the plant floor to optimally run an operation, most manufacturers are compelled to limit the amount and type of data to only the most strategic and/or problematic areas.

There are also legitimate concerns about how a direct interface to the plant would impact the bandwidth and throughput of the overall system. It's safe to say that current ERP implementations are designed to work around this limitation by making assumptions, calculations, and derivations, and to supplement these with manually entered data when it is available.

SCM

Among those aspects of SCM related to the plant floor, there's a critical dependence on manually entered data for event monitoring and management. It becomes a difficult and often impossible task to build a forward-looking model of your operation when there is only a limited and delayed view of the past.

Scheduling and Optimisation

Schedules are currently created and optimized based on large numbers of assumptions and calculations. The most significant of these include the actual availability of all required assets/resources/machines, the actual time required to complete a defined volume of production, scrap percentages, change-over or setup times, the actual running time required to complete a given production schedule as a fraction of the total time (incorporating the total time of all non-producing states).

There's obviously tremendous value in knowing in real time exactly what's being produced, when a specific production schedule was initiated on any particular machine, how much scrap and downtime is associated with each production run, and the overall efficiencies per product and per machine.

Maintenance issues

Most predictive and preventive maintenance is done today based on time and the manual aggregation of downtime sheets once a shift or once a day. By the time this data is aggregated and reported, (assuming that what was documented was accurate and unbiased), it's often too late to react. Critical to effectively deploying maintenance resources is clear and dynamic information about how every asset in a facility is performing.

Knowing the top 5 faults by frequency and duration for each machine/line/department by shift period/shift/day/week/month and by product/lot/batch/etc. is the key to proactively deploying your maintenance resources where they will make the biggest impact.

In addition, driving a maintenance system by time (as opposed to some combination of machine utilisation, cycle count and condition/threshold monitoring) is cost prohibitive; it results in doing more maintenance work than is actually required, and loss of production while those machines are being maintained.

Production Monitoring

Knowing dynamically how much you're producing and scrapping; knowing where your plant/department/line throughput constraint is; knowing how much inventory is tied up on the plant floor; knowing the sources of downtime by machine/operator/shift/line/department/week/month etc. - all these would seem to be critical to any quantitative decision-making process in the plant. Yet to collect and aggregate this type of information manually would require a tremendous amount of time and effort.

Enterprise manufacturing

The same issues and constraints that exist at the plant level exist at the enterprise level. In many cases a manufacturer will be producing the same or similar products in multiple plants, and the requirement exists to be able to quantitatively compare production statistics, efficiencies and overall Key Performance Indicators (KPIs) from plant to plant.

In an environment where plants often compete with each other, the incentive is there for the plants to make this data - or such of it as exists - as obscure or optimistic as possible. The plants need the flexibility and freedom to measure and analyse any way they choose, and corporate requirements should dictate only the specifics of a fixed number of measurements and KPIs as appropriate.

There's clearly a requirement for a manufacturing software solution that acts both as a manufacturing intelligence platform and as a bi-directional pipeline between the plant floor and various business systems.

The benefits of manufacturing intelligence

It would be required to collect, visualise, analyse and report on any plant floor data - both in real time and for any amount of history.

It would have to support hundreds of different types of users of every imaginable job function, each with the ability to create their own view of the plant floor - their own personal dashboard.

It would have to be off-the-shelf, so that it could be maintained with in-house, existing skill sets, and not require any software or database skills.

It would have to be 100 per cent web-deployed and server-side managed, so that the TCO would be feasible, regardless of the size of manufacturer or the size of the plant.

It would have to be mission-critical, guaranteeing data integrity along every step of the way, and designed to be running 24x7. It would have to allow the modification or addition of assets, machines or data points on the fly, without ever shutting down the collection engine.

It would have to interface simply and seamlessly with the devices on the plant floor and with the business systems that would benefit from the information it would provide.

So whether the aim is production monitoring, more effective scheduling and optimisation, improved throughput and efficiencies, dramatic quality improvements, reduced downtime, or part tracking and containment, the fundamental requirement is for one mission-critical software platform that can perform all of these functions seamlessly, thereby increasing production and profits.