Opportunities and risks of technology and analytics

What is the single most important component in industrial computing?

No, it isn’t the technology. It’s not the network, either. Believe it or not, it’s you!

Much manufacturing technology buzz centers around the industrial internet of things (IIoT), cloud computing, artificial intelligence, and analytics. Such technology’s value lies in its ability to help you discover and learn. Your ability to use your learned information lies in the depth of your understanding of your business model, manufacturing processes, and how different materials behave. It’s all just artifacts until you use your knowledge and the right mathematics to gather valuable and actionable information from your data. Properly planned and executed analysis can lead you to discovery and learning. Learning leads you to better decision-making. Be aware, though, that poorly executed analysis, whether bad assumptions or bad mathematics, can lead to costly, sometimes catastrophic decisions.

The Price of Inadequate Domain Knowledge

An often repeated story in the financial world is the case of the “Salad Oil Scanda”l of 1963. During that time, Allied Crude Vegetable Oil in New Jersey obtained millions of dollars in loans against its vast inventory of soybean oil. Ships would arrive at Allied’s docks with their holds supposedly full of salad oil to unload. In fact, the ships were full of water with several inches of oil floated on top. Auditors inspected the inventory, unaware that it was mostly water, and gave Allied the credibility to obtain millions in loans. Once the scam was discovered, it cost more than $150 million (more than $1 billion in today’s dollars) in losses to American Express, Bank of America, and others. American Express lost 50 percent of its stock value because of the scam.

Had the accountants taken the time to understand the industry, they would have learned that Allied’s alleged inventory exceeded the entire U.S. inventory of soybean oil at the time. They may also have verified inventory through means better than simply pulling a dipstick through water with oil floated on top. Once again, overconfidence, laziness, and inadequate knowledge lead to poor analysis and costly decisions.

Data and Technology: Tools to Enable Your Own Greatness

Data is just artifacts, whether it’s big or not. Cloud computing is just somebody else’s equipment and software. As a matter of fact, cloud computing is quite similar to the timesharing model companies used in the 1960s. The term cloud comes from the propensity of software designers to take shortcuts by displaying the internet as a cloud on their diagrams. The IIoT is a network of many of the things that generate data about your operation. It can all be a blessing or an overwhelming curse, unless you are prepared.

All the new tools to collect, accumulate, and share operational data can open new opportunities for you to discover, learn, and predict operational events. Technology can help you improve your operational, supply chain, and maintenance results. Making data useful, though, requires deep domain knowledge about your equipment, your material, and your operations. Without adequate knowledge of your business—the fundamentals of maintenance, sheet metal properties, and forming processes—you expose your company to some really bad decisions.

Data is generated by events and transactions. For our purposes, an event is something that happens, resulting in the generation of new data or a change to existing data. A transaction is an event that generally involves an exchange between two parties. Transactions include sales, purchases, contract execution, and product returns.

Analyzing data includes both the art and science of discovery and learning. As with any project, your success depends on your ability to plan and exercise your analysis. To create valuable and meaningful analytics, you should perform the following:

• Define your learning objective. Explain your goals for your analysis project. What do you want to learn? What problem do you want to remedy?
• Define your scope. To keep your project from growing beyond your budget and time, you must define the boundaries. Your scope may include such things as a named facility, a distinct subset of operational activities, or a particular grade of sheet metal.
• Learn your underlying process model. To understand what your data means and its operational context, you must understand the processes that generated your data and the relationships among those processes. Even if you’re analyzing condition monitoring, you should know the operational state of your equipment, normal expected conditions, the loads your equipment may be under, and the consequences of undesirable conditions.
• Define the information you need to accomplish your goal. Single events can sometimes generate large amounts of data. You may need to eliminate unnecessary data to avoid confusing your audience, to maintain your project scope, and to improve the efficiency of your analysis.
• Identify the equipment, transactions, and events that contribute the data you need. Eliminate unnecessary data.
• Define your approach to analysis. Will you need basic mathematics—totals, products, ratios, etc.? Will you need to understand statistical properties—averages, means, population distributions, probability of occurrence? Will your analysis be historical only or used as a predictive model?
• Verify your analysis. Confirm your approach. Be absolutely certain that your mathematics are solid. Confirm that sample size will support your desired confidence level and an acceptable margin of error. Make sure that your data is appropriate for supporting the right conclusion.
• Recommend actions. Once your analysis is complete, recommend remedial actions to the processes within your scope.
• Define how you will present your information to your audience. Prepare a presentation appropriate to the level of understanding of your audience.

Technology’s biggest advances include:

1. Affordable storage that allows you to collect massive amounts of data.
2. Exponential increases in processing power.
3. Increases in networking capability.
4. Improved algorithms that allow you to dig deeper into your data and respond to new knowledge.

Computers can’t think by themselves, nor develop their own understanding. That lives only in the realm of science fiction. Knowledge, understanding, and the ability to use discovery for good remains an innate human capability. You can take advantage of the opportunities technology offers. To do so, you must develop and support your own understanding of your business, your manufacturing environment, and the capabilities and limitations of the materials you use.