Normally we detect degradation from routine inspections or continuous condition monitoring until a component eventually fails an engineering specification – form, fit or function. With respect to engineering functional specifications, most components, assemblies, appliances and systems have multiple functional attributes and even more fault and failure modes. Failure of a component is when at least one critical attribute is continuously operating outside of an engineering form, fit or functional tolerance specification.

In prediction, one can predict engineering functional failure or non-failure (also known as false positives and the prime reason for no fault found (NFF).

The first of these is pretty obvious. A sensor or inspection indicates at least one critical engineering attribute is outside of a tolerance level and it truly is. The converse, non-failure, is when a sensor or inspection indicates an out of tolerance condition but it actually is not. This is called a false positive or Type I error. There is also what is known as a false negative or a Type II error.

So prediction identifies that a component is headed down its RCM P-F (interval between Potential failure and Functional failure) curve at some point in time before failure, given some level of predictive accuracy or it predicts and prevents no fault found. The bottom line value proposition of prediction is delivered in time. In the diagram, it is the time to respond prior to failure or non-failure without taking a revenue generating asset out of revenue generating service.

Dynamic diagnostics.
Traditionally, diagnostics has been considered a lagging or reactive capability – that is – engineering form, fit or function is already degraded, fault mode codes are being sensed or full failure has occurred. While in many cases this maturity level of diagnostics is still commonplace, especially when using static TSM (troubleshooting manual) or FIM (fault isolation manual) manuals, in the context of predictive maintenance and predictive analytics, diagnostics is becoming proactive. Customizable diagnostic reasoning engines that operate on a database of known equipment issues (i.e. a fault isolation) can capture all of the symptoms, causes and solutions for every known failure mode for a specific type, model, series or even individual piece of equipment. By dynamically generating decision logic, these proactive diagnostic systems identify the root causes of problems to guide users to the right prescriptive and corrective actions.

New generation commercial aircraft create up to 1 terabyte of operational data per flight from the plethora of sensors on board. Additionally, they have become software driven devices and there is an unintended consequence of hardware that doesn’t function correctly without the correct software, that being, diagnostics becomes unmanageable in a world of static CMS (content management system) enabled TSM/FIMs.

Jeff Immelt, CEO at General Electric succinctly sums up the new world order, “If you went to bed last night as an industrial company, you’re going to wake up this morning as a software and analytics company. The notion that there’s a huge separation between the industrial world and the world of digitization, analytics and software is over.”

And Mr. Immelt’s observation is at the center of much of the increasing complexity in diagnosing faults in aircraft systems and components, the constantly changing hardware, software configurations and the static nature of paper or even electronic content management systems.

The lack of predictive maintenance and advanced analytics capabilities adversely impacts fleet managers’, line mechanics’ and AOG (aircraft on ground) desk managers’ ability to support D-15 technical dispatch rate punctuality.

At the April MRO and OPS IT conference in Miami, Airbus presented their long range plan to incorporate predictive maintenance and advanced diagnostics capabilities into their OEM (original equipment manufacturer) specific solutions and we can be assured that Boeing, Bombardier, Embraer, GE, Pratt & Whitney and others will also build OEM asset specific tools. But even if an airline is a single fleet operator, unless they also perform all of their maintenance, repair and overhauls in house – I don’t know any who do – then OEM specific tools are no real enterprise solution. And of course the OEM-MRO ecosystem requires open standard interoperability across Information and Operations technologies and architecture for the entire industry to realize the value of everyone’s investments, something that mere data standards will not solve.

IoT operational and informational architectures
Predictive maintenance must be implemented in actual airlines’ current and future technology systems. IoT is commonly described as combining Information Technology (IT) and Operational Technology (OT). IT is exemplified by ERP (enterprise resource planning), SCM (supply chain management), CMS and MRO systems whereas OT includes the on board Central Maintenance Computer (CMC), engine control and avionics systems and even near board systems like diagnostics or health management systems. An aviation OT architecture might appear like the one below:

This data is being processed in flight and post flight, combined with human actions and content, and then combined with IT in order to predict, diagnose and prognosticate who should do what, where and when in order to optimize asset utilization, labor productivity and supply effectiveness. This requires a combined OT and IT integration. Because authoritative data and transactions move from an aerospace OEM’s Product Lifecycle Management (PLM) system thorough manufacturing execution (MES) to an operator’s Service Lifecycle Management (SLM) system, the ecosystem IT architecture looks like this:

To actually realize value from PdM capabilities, IT and OT systems will need to be integrated and matured. Doing that starts with an understanding of the state of these systems today and a plan on where to go moving forward. And because IoT is an ecosystem set of capabilities, they cannot be integrated or matured in a vacuum. If your aircraft or engine or component OEMs don’t enable OT sensing or if your MRO vendors don’t improve their MRO, CMS, digital and mobility capabilities, then building the best internal IT and OT will mean realizing less value from your investment thus increasing the costs for everyone in the aviation ecosystem.

Servitization: performance based business models
Probably the most disruptive element of IoT is how it enables servitization of ‘products’ or ‘devices’ and thus totally changes a company’s business models. It is interesting to note that, in the general media, the most often cited example of how IoT fundamentally changes a company’s business model via servitization is Rolls-Royce. We in aviation are very familiar with ‘Power by the Hour’ and how it bundles a ‘product’ or ‘device’ funding, purchase price, warranty, maintenance, service parts, etc… and denominates the price paid in what is actually being consumed – thrust. This is what is meant by ‘servitizing’ a ‘product’.

In IT this is commonly referred to as Software-as-a-Service (SaaS), Infrastructure-as-a-Service (IaaS) and Platform-as-a-Service (PaaS). Salesforce is a prime example of both a PaaS, where any software vendor can develop and host their solution, while also being a SaaS for their own CRM solution. And in order to optimize revenue and profits and market share, Salesforce PaaS hosts competitor’s CRM SaaS solutions. This is what is known as coopetition or the combination of cooperation and competition.

The greatest challenges for companies from this point on are not technological – transport protocols, data standards, security, connectivity, etc.; rather, they are centered on realigning their thinking, strategy and business models for servitization, a future that has already arrived.

There is no such thing as a product. Products only exist in the minds of the inept. Products only exist for the ‘spoilable’ service they deliver. Services and Service Lifecycle Management is where the check is cashed. Again, it’s much easier for us in aviation to understand these principals since an aircraft is really only a factory for creating available seat miles (ASM) for pax and available ton kilometers (ATK) for freight. Aviation metrics are all denominated in RASM (revenue per available seat mile), CASM (cost per available seat mile), LF (load factor), RPM (revenue passenger mile) and Yield.

Back to our Google Nest example. Even Honeywell Home Automation would probably, privately admit they were asleep at the switch (pun very much intended), they are arguably still asleep. While Honeywell has developed a similar ‘device’ or ‘product’ to Nest, that includes basic remote programming and control of a home’s HVAC, Nest is partnering with electricity and natural gas utilities and heating and air conditioning manufacturers to build a total home energy management offering, servitizing not only their own devices but HVAC’s devices too.

As many airlines have now experienced as they have brought new generation aircraft on board, OEMs are placing onerous constraints on their use of home grown and, or third party technologies. The battle over Operational Data and content has gotten worse. Counter to IT and OT trends in most other industries, with few exceptions, aerospace is becoming more OEM proprietary, non-standard, non-open.

GE invested $105M USD in Pivotal, a provider of cloud IaaS and analytics services. After such an investment, most companies would consider Pivotal a strategic competitive asset and try to lock out competitors. To the contrary, GE has opened their Predix platform to the world. Coopetition at its finest.

Moneyball, changing an unfair game
If you haven’t read Michael Lewis’ book, Moneŷball: The Art of Winning an Unfair Game, or at least seen the movie, I highly recommend it. It’s for anyone who loves baseball or operations research or just a good story about people and family.

Baseball is fundamentally unfair because of the revenue distribution and ability for big-market teams – the New York Yankees, Boston Red Sox and Chicago Cubs – to generate revenue that allows them to pay more for their players. That makes it nearly impossible for small-market teams such as the Kansas City Royals, Milwaukee Brewers and the Oakland Athletics (A’s) to compete. Of course you can say the same thing about business. Some are global multi-billion dollar behemoths with more free cash flow and capital than they know what to do with; and then the rest of us are in the small to medium enterprise domain.

Moneŷball examines how Billy Beane, the general manager of the Oakland A’s, used unconventional data and statistics to smartly and affordably assemble a competitive small-market baseball team in the early 2000s. Granted, there is nothing extraordinary about using statistics to win baseball games… or business for that matter. Every kid who collected baseball cards for the past 100 years knows the runs batted in (RBI) or earned run average (ERA) of their favorite players.

Statistician Bill James, a pioneer of sabermetics (the basis for Beane’s Moneŷball approach), took baseball statistical analysis and prediction to the next level when he began publishing books about it in 1970s and 80s. The revolutionary idea that James developed and Beane discovered was that the traditional metrics used by general managers, scouts and teams to forecast player value, for decades, did not actually result in team wins.

For example, instead of ERA, Moneŷball uses Peripheral ERA (PERA), the ‘expected’’ ERA taking into account park adjusted hits, walks, strike outs and home runs. Instead of RBIs, James developed Runs Created (RC) which forecasts the number of runs a hitter is ‘expected’  to contribute to a specific team taking into account the parks that team plays in.

Billy Bean explained the reason for seeking a new way of prediction; “We had to look for new knowledge, we had to question everything. We found great inefficiencies in how everyone was evaluating players. “

Beane broke the biases of his scouting staff, who had always done things their way, by ignoring qualitative opinions like age, appearance, personality and playing style as well as traditional quantitative metrics like batting average and RBIs and instead relied on often overlooked statistics like on base percentage and slugging percentage – numbers that were actually better indicators of a player’s ability to create runs. Because creating more runs than the competition wins games.

As he explained it, “It’s really about information and it’s about making probabilistic decisions – we couldn’t afford to invest something and not get a return. We couldn’t take risk, we had to look at things like an actuary looks at things and understand future decision risk. This is really about getting the most information to make good decisions. We were trying to rob some of the things that were going on in Wall Street and apply them to baseball.”

The strategy worked! The A’s reached the playoffs in 2002 and 2003 and were competitive with the New York Yankees, a team that spent $80+ million more on its roster’s payroll. With this success, the Moneŷball phenomenon transformed baseball…and also the world beyond.

Moneŷball is about reducing risk while increasing the expected value of actionable decisions through quantitative and predictive analysis of a lot of disparate data, more commonly referred to as ‘big data’.

Enter 100 billion sensors, 50 billion connected devices, a yottabyte of data and $14.4 trillion in economic impact.

The consumer and industrial Internet of Things are pontificated to change how businesses deliver everything. IoT is ushering in the biggest and most disparate increase in data seen since the invention of the internet, leading some to say ‘big data’ is the equivalent of ‘oil’ – a currency unto itself.

But you don’t run cars or factories or power plants on oil. Just like oil, big data must be processed into fuel for use; sensor and transactional data needs to be processed and transformed into actionable knowledge before systems and, or people can make decisions and respond in a timely manner.

When Rick Wysong was VP Engineering, Strategy & Planning at United Airlines, he once told me UA was filling Teradata data bases up with ‘tons’ of data and that, maybe six months after the fact, his engineers would figure out what they should have done six months earlier. Rick didn’t need more data, he needed better faster decision capabilities.

Accelerating sense and respond and improving accuracy and precision lies at the heart of value creation and value realization in IoT and PdM. Which brings us back to Will Greene; “How do I cash that check?” That simple question anchors every successful CxO’s decision making process. Realizing the value of industrial IoT begins with accurate and precise decision-making focused on the right criteria, outputs and outcomes throughout an organization. That’s the Moneŷball effect. So what does Moneŷball teach us about cashing the check in IoT?

• It teaches us that what is termed the Internet of Things is an equalizer for small to medium sized companies to compete and collaborate with ‘the big guys’;
• That there is sedentary inertia within organizations not to change thus requiring visionary and assertive leadership to overcome the innovator’s dilemma;
• That IoT applies to both planning (buying the right equipment, players, partners) and execution (operations of people, plant, equipment, marketing, sales);
• That merely connecting edge devices does not, in and of itself, realize value;
• That the ability to sense and gather data (big, small, or otherwise) creates potential value that must be realized across the Product and Service Lifecycles in addition to Customer Experience;
• That companies investing in IoT need to look at asset and operations performance levers and how sensing, monitoring, controlling, predicting, diagnosing, prescription, prognosis and autonomics improves decision making to change outcomes; and…
• That predictive analytics (predictive maintenance in industrial markets) delivers the lead-time required to respond to asset, operations and customer events in order to optimize return on assets, return on operations and return on investments.

So, what is the Internet of Things?
The Internet of Things is neither about the Internet nor about Things; it is about creating outcomes (baseball runs and wins or safe on time departure of aircraft).

It’s about connecting edge devices that sense conditions to systems that monitor, analyze and process stochastic events, automatically gathered by both machines and humans, to create value from autonomic manipulation of data into knowledge for the purpose of making timely actionable decisions in order to realize value for consumers of spoilable services.

Contributor’s Details

Michael Wm. Denis

Michael Wm. Denis is a renowned author, speaker and independent consultant providing strategy, business model, organization, marketing, sales, operations and technology advisory services to manufacturers, operators, maintainers and technology vendors focused on optimizing the service lifecycle of complex capital assets. Among other clients, Michael is advising several predictive maintenance solutions vendors’ strategy and business development in the aerospace, aviation and defence markets. www.slm.aero

CaseBank Technologies is a pioneer and leader in delivering diagnostic solutions for complex capital assets that improve first time fix rates, labour productivity and equipment reliability. CaseBank’s Spotlight®, ChronicX® and Diagnostics Data AnalyserTM incorporate actual feedback from field service technicians to accelerate future corrective actions. CaseBank guides the fault isolation processes and procedures to optimize the efficiency and consistency of product support. CaseBank’s solutions are used by maintenance organizations in a variety of industries, including aviation, defence, automotive, rail, industrial equipment, and high tech electronics.