The Future of Information in Intelligent Devices

By Matt Miller, OSIsoft Inc. – Ann Arbor, MI [email protected]

It’s almost a cliché to say it:  since the invention of the computer and the start of the technology age we have been on a high-tech roller coaster ride, one that has no apparent end.  The change at hand has filtered from large business systems through the Internet technology into our homes and even into the appliances that fill them. 

Now, these once simple devices are going through a metamorphic change as (a) they become more intelligent and  (b) pervasive networking becomes the norm in business, private homes and institutions.  Although refrigerators and washing machines won’t take over the world, intelligent devices will fuel the productivity thrust during the coming decade.  In a recent analysis of the ‘90s, Federal Reserve Chairman Allen Greenspan commented that its prosperous economy was driven by an increase in individual productivity of more than 3 times! This was due to the widespread adoption of the PC; the next step in both business and private life will be always-on, networked embedded systems, many of them extremely small and portable.

The HAL Syndrome

It’s difficult—or better yet, impossible—to predict the future from extending present trends.  Take some examples.

  • In the ‘30s, autos traveled at 70 mph on dangerous roads.  By the ‘70s cars had slowed to 55 mph on freeways that could have allowed much higher speeds.

  • U.S. commercial aviation, from its inception until 1958, saw a steady increase in aircraft speed.  In1958, seeing that the new Boeing 707 jetliner could fly at nearly 600 mph, one might have predicted that forty years later planes would fly three to ten times faster. But except for the Concorde, commercial aircraft today fly just a few percent faster than they did in 1958. Few could foresee that the US would never allow sonic booms over its landmass.

  • Since the vacuum tube was invented in the mid-1870s, much effort went into making it more compact and power efficient. Tiny low-power tubes came out in the late ‘40s; and a knowledgeable engineer might have predicted that one day soon, vacuum tubes would be the size of a pencil eraser and would consume just a fraction of a watt.  Instead, in 1947, Bell Labs invented the transistor.

To those of us who can recall the computer HAL from the epic film 2001 A Space Odyssey, HAL’s intelligence seemed to embody the inevitable evolution of the computer. Instead, reality moved in an entirely different direction.

Rather than ever more centralized, large-scale intelligence, technology has evolved towards highly distributed, intelligent networked devices.  A major reason is the cost benefit ratio of intelligence.  We can now achieve reasonable computing power for around $25—but a machine like HAL would cost millions or even billions!  The result is that small computers and micro controllers are being embedded in even the simplest devices to add value and provide the intelligence to make our life easier and our work more productive.

An easy way to see the market’s thrust toward ever smaller intelligent devices is to examine the relative growth of personal digital assistants (PDA’s) such as the Palm compared to traditional PCs.  The PDA market has been on a rapid growth path since the mid-1990s. PDA sales surpassed 5.5M units in the USA in 2000 and yearly sales are projected to reach 21M in 2007. U.S. PDA revenue in 2000 reached nearly $2B and will grow to $7.6B in 2007. In 2000 the volume of PDA units was at 12% of PC units, but this will increase to nearly 37% in 2007 (although the relative value will not change so drastically due to the difference in price between a PC and a PDA).  The technology is evolving rapidly toward the smaller, dedicated functionality of the PDA, especially as it becomes linked more and more with communications technology such as cellular phones.

Almost every household device today embodies this change. Your new microwave oven, coffee maker, refrigerator, washer now come with the intelligence to monitor their own operation, detect potential problems and provide usage information that helps you make better decisions. As the marketers in Redmond tell us, this increases our “experience” with the device. 

Recently I had a new water softener installed in my home.  To my surprise this appliance was quite intelligent.  With an onboard computer system it monitors all aspects of my water usage:  flow rate, total usage, average usage, average flow rate, salt usage, % available softened water left, estimated time to the next charge, and many other pieces of data that I have yet to explore.  One of the unit’s most impressive features is its ability to send me an email when it needs additional salt.  All this intelligence came “free” with the mid-price unit.  I’m glad we didn’t purchase the high-end unit: it might wake me in the middle of the night to remind me about overdue maintenance!

Such advances present a growing challenge for manufacturers of large industrial machinery that costs tens or hundreds of thousands of dollars: users who have become accustomed to an almost magical intelligence in simple devices at home are coming to expect the same and more from these impressive machines.  It also presents a challenge for users who make high-volume industrial equipment such as sensors, control devices and embedded instrumentation used throughout factories, plants, mines and refineries.  If their microwave at home can call for help when it’s broken, operators will expect nothing less from a pump or valve costing ten times as much—or from a front wall coal mining machine costing hundreds of thousands as much.

So what’s next?

Benefits of networked, intelligent devices will be many.  But it is important to keep in mind that there will be two separate groups of beneficiaries—users of the intelligent devices, and their manufacturer—and the benefits they derive will not always overlap.  Obviously in some cases they will coincide, for example when intelligence allows the manufacturer to provide better customer service or improved maintenance.  In other cases, such as extracting usage information in order to make a better product that the user might never purchase, these benefits will accrue primarily to the manufacturer.  

 

 

Function

Benefit

User

More access to data using a Data Historian

Improved analytical powers, system optimization for high productivity, world class performance

 

Preventive maintenance

Less downtime, higher productivity

 

Predictive metrics

Less waste during setup, greater flexibility in using multiple sources of supply

 

Time stamped data

Necessary for optimizing large and complex systems

 

Knowledge integration

Real time information makes it possible to manage to business strategies on a day-by-day basis

Manufacturer

Remote diagnostics

Improved customer satisfaction, reduced service requirements

 

Ability to collect data about customer use

Improved market research and design procedure, reduced warranty and liability exposure

 

Product differentiation

Intelligence augments the core function of the device; has significant PR value; will eventually become necessary to stay in market

 

More data

Providing historical information for simple machines is relatively easy.  In the case of a water softener, to calculate the average water usage is a simple calculation that can be represented in a single text phrase.  The same process for a citywide power grid monitoring system requires analysis of thousands of points simultaneously.  As machines get more complex so do tools required for analyzing their data. For this purpose a software tool called an Historian can be used to retain the process details of all the points in their original fidelity.   Historians have been deployed by progressive manufactures all over the globe to achieve world-class operational performance.

All the data, all the time

As systems are deployed to provide key process insights and allow users to better understand their processes and improve them, data gets refined.  This refinement generates a need for even greater access to similar data from other machines that support the process.  Although much of this data may not be gathered at the highest level, the ability to “drill down” when additional information is needed will provide an almost limitless resource for making process improvements.

In certain cases the fact that so much data can be stored in the backend of the device may provide legal protection to the manufacturer.  It will now be more possible than ever to prove when a malfunction took place—and in many cases why.  This will simplify processing warranty claims and even settling issues of legal liability.  What insurers call the ‘moral hazard’ will be reduced.

Continuous Improvement Cycles

All this data is an endless source of identifying potential improvements in a process.  In nearly all cases, the act of getting the information is the most time consuming and costly part of the continuous improvement cycle.  Having all the data readily available from the beginning compresses the cycles in the improvement process enabling agility and responsiveness to market pressures and better implementations.

Highly distributed networks

Born from PC technology and proliferated by the latest software development tools, the ability to connect intelligent devices has become a necessity.  This poses a dilemma for most equipment manufacturers.  Providing an “open platform” system exposes them to support problems, increased support costs and maybe even unhappy customers.  If my water softener took two weeks of customer support to use, the company would never make any money and I would never recommend them to others. 

On the other hand providing a secure method of analyzing machine metrics can lead to higher user satisfaction (and more sales).   Microsoft has done much to standardize the way systems communicate by providing layers of network infrastructures that let systems share data in a secured fashion.  This will foster the ability of systems to share information with each other while limiting the access via standard security measures.

Predictive metrics & methods

A major benefit of local historical data is that it enables a user to optimize and tune the local process for better results and reduced waste through better understanding of process fundamentals.  This information can help understand the effect of process variables such as differences in materials, environment, personnel and machine wear.  Once such differences are analyzed they can create predictive metrics to help prevent the problem recurring. 

Take the example of a laminating machine.  The user gets materials from multiple suppliers based on availability, price, quality and terms.  Slight adjustments for each vendor’s material must be made in the “press” operation of the machine to laminate the materials correctly.  An Historian on the machine after a few runs with materials from Vendor A would show that the pressure used in these batches had been within certain ranges.  After a few more runs the user may be able to provide a statistical prediction of the pressure for the upcoming run based on the material mix of the product, eliminating defective product in the batch during setup.

Knowledge integration

Collecting this knowledge is the first step.  Sharing it with other systems such as MRP and ERP is another.  The smooth flow of information in real time from the plant floor up to the business management systems used in the IT department will make the CFO’s job that much simpler; it may be that quarterly reports will be replaced by daily reports.

Historical Data in “real-time”

For many applications the information is only relevant for a specific time.  Like the price of a stock, without knowing when the stock price was it would be of little use.  Historians provide a key role in the storage of time series data.  Time stamped data makes it possible to determine exactly what happened when, an invaluable tool when examining the operation of a large and complex system.

The ability to collect information about customer use

A sensitive area is collecting information about customers’ use patterns.  However, pervasive networking may provide manufacturers and their sales and support chain with useful marketing and product information via a live connection into the factory floor. Data received from sensors within the equipment—both large machinery and smaller, embedded devices—will lessen the need to conduct customer surveys during future product planning and design. Devices could be used for collecting and reporting usage information.

The prudent supplier will ensure that they have buy-in and procedures for protecting the user’s proprietary information before going ahead. It must be borne in mind that visionary adopters are the least likely to want to exchange information on their usage for certain privileges with the supplier since they will be seeking a competitive advantage in adopting a proprietary improvement in their manufacturing or processing methods.

Remote diagnostics

A major opportunity arises in improving the supplier’s service quality. The ability to service an intelligent device remotely and head off technical problems is a clear practical advantage. A networked device can enable diagnostic data to be sent automatically to a technical support center and components can be detected before they fail. Both warranty management and maintenance firms can be fed machine status information and the manufacturer gains far more detailed information about a machine’s internal operations in the field than a customer survey would ever yield.

Combining the data from all the devices in the field in this way will allow analysis of operational parameter patterns that correlate to failures – such as calls to the technical support center. Once determined, these patterns could then be used for preventative maintenance service offerings. Together, these factors make an individual machine a much more attractive proposition for the buyer whilst augmenting the supplier’s revenues and operating efficiencies.

Product differentiation       

Perhaps more important than anything else is the boost that the network dimension gives to the device in terms of brand differentiation. Initially, like most new technologies, intelligent devices are moving from visionary purchasers out to all categories of users. Economic benefits will need to be promoted as products spread to the mainstream market. In the near future though, the PR value brought by the manufacturer’s association with cutting-edge technology concepts is likely to be a significant value. As regards product differentiation, the integration with enterprise-wide SCADA/IT systems sets a product apart, providing the user with services beyond its primary function and more importantly, augmenting the core functions of the product.

…And what is to be done?

In response to these advances in intelligent devices and related changes in networking, sensor and integration technologies, combined with lower prices, machinery manufacturers have begun to expand their products to offer a range of Internet-based services. As with the automotive industry, forward-looking machinery manufacturers have started to develop network capability and integrate Internet-based services with their products, transforming them into platforms for service delivery.

As a result of these new technology waves, closer customer relationships will be formed throughout the product’s lifetime. ‘Service’ usually means ‘repair’ or ‘maintenance’ today, but means something quite different with the opportunities presented by pervasive networking. The range of services – and revenue streams – is virtually limitless.

Suppliers need to be considering how the factory will be run in the future and the implications of this fundamental transformation to their business models. By taking advantage of falling communication hardware costs, networked sensors and always-on Internet connectivity, new services can be created and product differentiation can be achieved. The Internet is evolving to interconnect all network-attached devices, effectively becoming the largest data acquisition and control network. Some key enabling factors driving this are the move toward always-on broadband connections, the availability of inexpensive wireless networking, both Wide Area Network (WAN) and Local Area Network (LAN) and falling cost of network connection hardware.

Suppliers need to be aware of the importance of integrating pervasive networking and intelligent devices into factories. Machinery and equipment manufacturers need to change their existing business model and internal structure to deal with the new, ongoing relationship with the customer. This growth in the importance and value of customer relationship is prompting a need to form partnerships with Information Technology (IT) or Internet services suppliers, application service providers and other industries. As a result, manufacturers need to change their concept of Customer Relationship Management (CRM) from one that effectively ends with the sale to one that promotes an ongoing involvement with the customer and builds on continued service to develop valuable, revenue-generating relationships. In addition, manufacturers will need to re-evaluate their corporate governance and decision-making control in this new environment and their supply chain strategies given the increased high-technology content in their products.

****

Matthew Miller is the OEM sales and Marketing manager for OEM products at OSIsoft, whose RtPM Real Time Process Management  platform provides the ability to help people capture, analyze, distribute and visualize real-time information for continuous process improvement at thousands of sites worldwide.  Matt has 20 years experience in embedded hardware and software for the industrial automation industry, specializing in visualization products.  He is currently focused on OSIsoft's ECHO embedded historian product (www.echohistorian.com) which is used to store automatically collected data in a wide range of industries.  You can reach Matt at 734-944-2484 or [email protected].