Readiness is an age-old focus for militaries. As budget pressures continue to build, digital resources that are now maturing—data science, cloud computing, and AI—can help the US military be more efficiently prepared.
From Pericles to Patton, military leaders have likely always wondered if they had enough soldiers, spears, or tanks in fighting condition. In today’s environment of proliferating threats, military readiness is once again in the spotlight. US Secretary of Defense Jim Mattis has made readiness a priority, while across the Atlantic, British, German, and French defense leaders are also formulating plans to increase the low availability of combat aircraft, ground vehicles, ships, and personnel.1 As militaries across the globe look to maximize the value of every dollar in their budgets, it seems time to reframe the debate on readiness with a new lens that leverages technologies, data science, cloud computing, and artificial intelligence (AI) capabilities unavailable just a few years ago.
Explore the government and public services collection
Subscribe to receive government and public services content
Readiness can inspire quite a lot of contention.2 Readiness has been conceptualized as both operational (a unit-level measure) and organizational (a strategic perspective). The concept of the full readiness spectrum reconciles this contention, by incorporating both the operational and strategic lenses of readiness into a comprehensive view. At the narrowest level, readiness is the training and health of a unit and the maintenance status of its equipment. At its most expansive, readiness is “the ability of military forces to fight and meet the demands of assigned missions.”3 The full readiness spectrum spans from this most narrow point all the way through the most expansive, providing decision-makers not only with high-level views needed for strategic decisions, but also the detail needed to make tactical improvements. Thinking about this full readiness spectrum sparks an important set of questions: Ready for what? Ready by when? Ready with what combination of forces?4
Readiness ultimately rests on one core question: Are we prepared to win the next fight? No single question might better define a military’s mission than whether it can muster sufficient forces, anywhere, at any time to defeat an adversary. But it is also a difficult question to answer: It hinges on a huge number of inputs and assessments that span from the tactical to the strategic ends of the spectrum, each of which can be challenging to gather and analyze. To help simplify the problem, this core question can be broken into three questions that decision-makers should understand:
Current readiness reporting systems often struggle to fully address each of these characteristics, but a new approach built on emerging technologies, including sensors, wearables, autonomics, and machine learning, can provide defense organizations with greater insight across the full readiness spectrum than they have ever had. Together, these technologies are enabling a shift in how industries operate. That shift, called Industry 4.0, is already enabling the private sector to prototype rapidly, build products faster, optimize operations, improve maintenance, and manage smarter supply chains. But to understand the full value these technologies bring to readiness, we first examine some challenges in the current system.
Current approaches to readiness measurement limit the visibility senior leaders have into the real-time, ground-truth status of their forces. That’s because readiness in its current form does not always accurately depict the true state of forces—instead, it’s a model. Current readiness metrics attempt to represent the state of the forces relying on a set of proxy inputs, but this system has two inherent weaknesses: First, reporting systems are fundamentally backward-looking; second, they provide decision-makers with only high-level summaries, not underlying data.
The current system, in effect, documents the results of resource decisions made months or even years prior, and by summarizing all the data together, tends to lose critical information about the root causes of readiness problems. In a reporting-based readiness system such as that used by the US Department of Defense, only net status reports—not the authoritative data sources—are forwarded to leadership.5 The result is that leaders are sometimes making decisions based on information that is only as deep as a single slide deck. What is needed is a system that can provide broad trends to support the decision-making of senior leaders but still retain the granularity of the underlying data.
Truly digital organizations recognize this, and they rely less and less on reporting alone to evaluate projects, performance, and maintenance.6 Instead, they use real-time data streams taken from real-world measurements. Every employee check-in, every rotation of a connected turbine, and every heartbeat sensed by a wearable device can generate meaningful data. Individuals can be compared with others and with their own histories. This new wealth of data means organizations can spend less time developing models of the world and start looking directly at the real world.
A new approach to readiness based on real-world data may seem almost impossible to implement. After all, finding the location and status of every piece of equipment and every service member seems a nearly insurmountable challenge. However, this large task can be made simpler by breaking it up into smaller and more manageable problems where existing solutions can help. In fact, portions of such a solution are already at work in many commercial companies using Industry 4.0. (See the sidebar “What is Industry 4.0?”)
The concept of Industry 4.0 extends digital connectivity into the physical world. Industry 4.0 technologies combine digital information from many different physical and digital sources and locations, including the Internet of Things (IoT) and analytics, additive manufacturing, robotics, high-performance computing, artificial intelligence and cognitive technologies, advanced materials, and augmented reality. These data sources come together to improve operations in an ongoing cycle known as the physical-to-digital-to-physical (PDP) loop (figure 1).
Throughout this cycle, real-time access to data and intelligence is driven by the continuous and cyclical flow of information and actions between the physical and digital worlds. Many organizations already have some portions of the PDP loop in place, namely, the physical-to-digital (sensors providing data about key assets), and digital-to-digital (digital analytics tools) processes. However, it is the leap from digital back to physical—from connected, digital technologies to action in the physical world—that constitutes the essence of Industry 4.0.7
For further information, see Forces of change: Industry 4.08 and Industry 4.0 and manufacturing ecosystems: Exploring the world of connected enterprises9.
The core of Industry 4.0 involves using digital information about the physical world to improve decisions and actions taken in the physical world. With more accurate information about the physical world, organizations can make better decisions, faster.
No company faces the same daily challenges as the military, so merely importing solutions wholesale from the commercial world will never work. However, the core issue of readiness is similar in nature to the business challenges faced by many companies. Therefore, the approaches and technologies of industry, when recombined in new ways, can help to solve similar challenges for the military. In this way, a manufacturing company struggling with efficiency could look very much like a military organization struggling to determine its readiness. Both need to know the location of their assets, the condition of key components, and whether the current status of equipment can meet the current demand from orders. To put that in the terms of the three readiness questions, both militaries and companies need to know the need, know the assets, and know the best action to take. For the military, this means that the three questions of readiness each require information of a certain type:
Each of these information needs can be addressed by different solutions, some of which are already at work in commercial plants and warehouses around the world (see figure 2).
Knowing what capabilities are needed for a mission may seem to be the simplest of all readiness issues for the military. After all, the units and capabilities needed for key missions are typically set out long in advance in strategic plans at every level of command. But there can be a large difference between military operations as planned and military operations as successfully executed. After all, as the old adage goes, “No plan survives first contact.” Therefore, to truly understand how ready a force is to execute a mission, militaries should not only look at what the plan calls for but also what has been done in the past in real-world missions.
This is much more than just looking at the order of battle or lists of units. Rather, it is about understanding how different parameters of a mission place demands on the forces that will execute that mission. Everything from the task, to the terrain, to the weather, to the particular enemy being faced can change the makeup and skills needed to execute that mission. Operating at high altitude? You will need more helicopters due to reduced load in the thin air. Tasked with a foreign internal defense mission? You will need junior personnel with local language skills. The huge variety of different factors make predicting future force requirements exceedingly difficult.
Here, machine learning can step in to provide an answer and ease the workload. Today, machine learning is all around us performing tasks from recognizing hand-written digits to recognizing patterns of credit card fraud. A machine learning tool using neural networks or another approach can be trained to associate the various mission parameters (e.g. task, enemy size, terrain, weather, time, and so on) with the size and capabilities of the force needed to accomplish them. (To read more about how machine learning can help planning an operation, see the sidebar “Industry 4.0 in the real world: Machine learning in supply chains.”) Over time, the neural network will become trained enough to anticipate the need for wholly new mission parameters it has never seen before. In other words, the machine learning tool will be able to reliably suggest the required capabilities for future missions. But most importantly, these suggestions will not be based on assumptions or best guesses; they will be based on hard-won lessons from real-world missions (see figure 3).
Restaurants may seem like a familiar, simple business, but many feature complex supply chains with planning and actions reminiscent of military operations.10 Ingredients must be delivered regularly so that they are fresh, but all of that depends on the rate of use, which depends on how often customers order different dishes—something outside the control of the restaurant. Traditionally, this was done manually by tallying up how much of each ingredient had been used and guessing when it would need to be reordered. This could lead to over-ordering and waste, or ordering too late and not being able to fulfill customers’ orders.
One restaurant chain decided that there was a better way to manage its supply chain. The restaurant chain used machine learning and AI tools to analyze point-of-sale data. Soon, the system was able to “learn” how much of each ingredient was used for each type of order. Rather than forcing planners to predict based on the information they had on hand at the moment, the restaurant chain used a seamless flow of real-time data to begin to sense, anticipate, and even forecast demand. This improved forecast accuracy by 25 percent, reduced the workload on staff, and achieved a 99 percent in-stock rate for each restaurant.11 Even more importantly, the benefits cascaded down the supply chain, with every supplier being able to better plan. The result was less waste, timelier deliveries, and the right dish on the customer’s plate—every time.
Better knowledge, better predictions, and better performance: a lesson equally applicable in the restaurant as it is to the military. In fact, the US Navy is using similar machine learning tools to prioritize and route the huge volumes of message traffic that ships and submarines receive.12
While a complete picture of real-world mission needs can ease planning, the right capabilities still need to be available to carry out those plans. This challenge goes far beyond any patch chart or doctrinal table of organization. It gets at the heart of the factors that make military units successful. Some of those, like having the right equipment, are easy to see and measure, while others, like having local knowledge or experienced leadership, can be more difficult to observe. A real-world, real-time picture of the location and status of personnel, equipment, and infrastructure down to the individual level can help address this challenge.
After all, it doesn’t matter if a unit is designated with a regional specialty in West Africa; it matters if the squad leader can speak French. In fact, this more granular picture of capabilities can supercharge the benefits from task organization that the military already enjoys. Today, task organization tends to exist only at the unit level, blending different units with different capabilities to create a task force tailored to a mission. However, with a more granular picture of the equipment, personnel, and infrastructure, commanders could combine individuals with the right skills and experience with equipment in the right place at the right time to create more capable formations in shorter time. Rather than just seeing red or green stoplight charts, a commander could see that a helicopter may not be ready for its squadron’s upcoming combat deployment due to inoperable radar warning system, but could support a disaster relief mission with a different unit.
Similarly, an individual soldier may seem “non-deployable” due to upcoming paternity leave, but a more detailed picture of his skills and availability may show that his language skills may make him an ideal fit for a short-term deployment to support a foreign exercise (see figure 4). This more granular picture requires real-time information about the status of equipment, personnel, and infrastructure:
Technologies such as augmented reality can extend this data advantage even further by allowing the perfect expert for any situation to be virtually present. For example, some airlines are already using augmented reality to connect maintenance personnel at small airports with experts so that they can make complicated fixes immediately instead of waiting for experts to fly in and creating costly delays.13 For the military, this could mean that if you know the capabilities of each service member, critically needed knowledge about anything from repairing a broken landing gear in the field to translating Nepalese can be only a headset away.
Simply gathering this data is not the end, however. Different forms and types of data should be stored in such a way that they can be analyzed together to allow both the vehicle-level mechanic and the corps-level general to discover insights relevant to them. This capability typically requires some form of big data storage that can support parallel and cross-cutting analysis—which may be neither as new nor as daunting as the technical jargon may seem. Predictive maintenance and IoT solutions in both commercial industry and the military have been using similar data stores for years, many of which are even open source.14 (To read more about how IoT and predictive maintenance are already helping organizations, see the sidebar “Industry 4.0 in the real world: IoT and predictive maintenance.”)
The challenge with traditional maintenance strategies is that it can be difficult to predict what will happen with a given machine. Therefore, companies typically must stock a wide variety of different spares to be prepared for any failure. However, IoT-based sensors and predictive maintenance algorithms can offer a more efficient solution by providing insight into the condition and status inside each piece of machinery.15
Italian train operator Trenitalia was looking for exactly such a solution when it faced performance penalties for schedule delays created by unexpected maintenance on its trains.16 To address the problem, Trenitalia added hundreds of onboard sensors on 1,500 locomotives. Data from those sensors was then transmitted to private cloud storage in near-real time, where diagnostic analytics provided advance warning of the failure of parts such as brake pads. With such data, Trenitalia was able to maximize the brake pads’ useful life while reducing unexpected failures. With these sensors and predictive maintenance algorithms, Trenitalia was able to decrease downtime by 5–8 percent and reduce its annual maintenance spend by an estimated 8–10 percent, saving about $100 million per year.17
While the maintenance demands of a commercial business with relatively predictable schedules are different than the hard-wearing, anytime, anywhere demands on military platforms, the same predictive maintenance techniques are already being used by some militaries around the globe.18 For example, the US Army is using predictive maintenance on a portion of its fleet of Stryker combat vehicles using more than 5 billion data points from onboard sensors.19
Having real-world, real-time information about equipment, infrastructure, and personnel would not only provide leaders with better information to support their decisions, but it could also support decisions up and down the chain of command: The local mechanics can see the state of their tanks, the commanders can see the state of all their tanks, and service leadership can see the state of all ground forces or the state of individual tanks. The unfiltered data allows leaders to zoom in and out, encountering both the granularity and the broad, emergent trends of their readiness posture (see figure 5).
The local mechanics can see the state of their tanks, the commanders can see the state of all their tanks, and service leadership can see the state of all ground forces or the state of individual tanks.
With a real-world picture of mission needs and a real-time picture of available capabilities, militaries can evaluate their preparedness in detail, potentially for the first time. But that is not the end of the story. Even more important is the question of how to improve that preparedness. When there is inevitably a gap between required and available capabilities, planners need to know where future investments should be directed to remedy that gap. Is the problem a lack of specially certified personnel, a shortage in a type of aircraft, or even a need to have more qualified units in an insertion technique?
What is typically needed is some sort of enterprise decision support capability. Luckily, machine learning techniques can again be applied here to answer those questions and ultimately ensure that the joint force is more capable, lethal, and affordable than before. By combining the real-world and real-time pictures of force capabilities, planners can create what is essentially a “digital twin” of the force.20 This digital, data-driven picture of the force allows leaders to conduct detailed scenario planning. They can test the force against different types of missions and combinations of missions in ways that would be impossible to recreate in real life. Moreover, by seeing how the force stacks up against these various mission requirements, planners can understand where gaps in capability exist and where the next dollar of investment should be spent to best fill those capability gaps—whether through additional capacity, greater training, or new materiel (see figure 6).
Take the readiness of an F/A-18D squadron as an example. If the squadron is experiencing readiness problems because of maintenance issues, solutions could range from decreasing operating tempo to increasing maintenance personnel, to expanding depot-level maintenance capabilities, to replacing the aircraft with a newer type. Today, planners would simply have to decide among these various solutions based on their understandings of the cost and efficacy of each. However, a machine learning tool applied to the “digital twin” of the force could draw on reams of data from history to highlight exactly which solution is likely to get the most jets in the air at the lowest cost. (To read more about how digital twins are being used by advanced organizations, see the sidebar “Industry 4.0 in the real world: Digital twin.”)
The collection of data and technologies described in figure 6 is what the industry would call a “digital twin.” A digital twin is an evolving digital profile of a physical object or process that helps optimize business performance. Digital twins start with complex models of the object or process they wish to mirror and then update them over time with data streaming from the real-world thing as it operates.21
The digital twin can mirror a physical object such as a particular jet engine or an entire process such as what is happening on a factory floor. Most importantly, these twins are not just for show but can be tested in ways that their physical counterparts cannot. For example, Formula 1 racing teams have digital twins of both their cars and the tracks they race on. They use these twins to run complex simulations of race strategies, often with the drivers in the car, so that on race day, they know exactly what tires to use, when to stop, and all the other settings that will give them the best chance of a win.22 Just like race teams, the military has long sought to prepare for the future with war-gaming and simulation. However, the ability to make those simulations more accurate via the use of a digital twin is a game changer and can produce a more lethal force at a lower cost than ever before thought possible.
The move from a reporting-centric picture of readiness to one that hinges on real-time, real-world information would represent a large departure for almost all militaries. However, the good news is that it seems to be a feasible change. Not only are many of the core technologies already proven in the commercial world, militaries likely have much of the needed data already. A digital transformation such as this one can often begin with the data already on hand without any need for massive investment in new technology or a bulky new IT system.23
In fact, some parts of the military are doing just that already. The US Air Force recently began using data that had been generated for years by C-5, C-130J, and B-1 aircraft to feed predictive maintenance algorithms.24 The same approach can be used for data that militaries already capture, from airfield statuses to daily personnel reports. All of this data has a value, and tapping that value is the key to understanding readiness in a new way.
Naturally, any transformation of a large bureaucracy can be challenging. However, hard-won lessons from other government and commercial modernization efforts can help smooth a military’s transition to a real-world, real-time readiness picture:25
At the end of the day, military advantage does not typically come from new technologies but rather from how technologies, processes, and people all work together.26 Industry 4.0 for readiness is about redefining those relationships. Armed with a few key principles, decision-makers can implement a plan to unleash the data in their forces and enhance readiness. By distributing proven technologies in new configurations throughout the military, leaders could soon have a full, real-time, real-world picture of their forces that only a few years ago would have seemed like science fiction.