Predictive maintenance 4.0 | Venturus

Predictive maintenance 4.0

Share on facebook
Share on linkedin
Share on whatsapp

Predictive maintenance consists of using machine condition monitoring to evaluate possible maintenance interventions as assertively as possible. With this, the goal is to anticipate possible problems, plan maintenance actions and perform those actions at the exact moment of need. To know more about the different types of maintenance, check out our article on Industrial Maintenance Techniques.

Sensor-based technologies are required to perform predictive maintenance. The most common are:

  • Vibration sensors — Velocity [mm/s] or acceleration [Gs] with spectral analysis;
  • Non-contact radial and axial displacement sensors (proximeters);
  • Temperature sensors (bearings and frame);
  • Thermal cameras;
  • Electric current sensors with spectral analysis;
  • Partial discharge sensors in electric motors;
  • Pressure sensors for compressors;
  • Linear expansion sensors for turbines.

From the reading of these sensors, it is possible to extract valuable data on the operation of the equipment. Techniques such as vibration analysis allow us to understand what is the defect being developed in the equipment — such as bearings, gears, unbalance, misalignments of shafts and pulleys, clearances, lubrication, among others. To know a little more about how this technique works, you can access our article on vibration analysis.

There are different ways to obtain data from equipment for predictive maintenance. One of them is route-based predictive maintenance, in which analysts visit equipment locations and collect data with portable devices, such as vibration analyzers with sensors, thermal imaging cameras, ultrasound equipment, among others. Another technique is online predictive maintenance, in which sensors are permanently installed on equipment and constantly send data to computerized systems.

 

Periodic predictive maintenance [offline]

Periodic prediction maintenance depends on a qualified professional to collect data periodically. Inspection routes are generated according to the equipment’s level of criticalness and its location in the production plant. Some equipment is monitored monthly, while others are monitored weekly or even daily for more critical equipment.

Professionals must be qualified and the equipment is specialized for industrial use. This raises the costs to implement this type of maintenance. On the other hand, large industries use this technique due to the high return on investment, since it has great assertiveness in predicting failures, as well as avoiding breaks and stops in the production line.

 

Permanent predictive maintenance [online]

Online predictive maintenance is a technique best applied to critical equipment in the event of failure. It consists in the installation of permanent sensors on the equipment, which collect data and send to electronic systems in real or near real time.

These systems process the data and generate information to be analyzed by qualified professionals, who perform interpretations of vibration spectra and waveforms, temperature and process variables.

Some types of equipment, such as robots, machining centers CNCs (Computerized Numerical Controls) and electric motors may have sensors pre-installed by the manufacturers themselves. In industry, equipment are integrated into automation systems through industrial protocols and can be analyzed by the maintenance team in expert software.

 

Predictive maintenance 4.0: IIoT + artificial intelligence

Industry 4.0 techniques and technologies can also be used in predictive maintenance, such as IIoT (Industrial Internet of Things) and Artificial Intelligence.

The cost of acquiring specialized sensors and hardware, cabling and electromechanical infrastructure, implementation, commissioning and start-up and maintenance of these systems is high, considering the need for skilled personnel to perform the necessary steps.

Therefore, in a project feasibility assessment, only large machines — such as injection pumps on oil platforms, large turbogenerators, long-distance belt conveyors in mining, that is, large machinery and equipment, with a high impact on production and safety — receive this type of system.

For a few years, several factors have been contributing to this maintenance technique being applied to an increasing number of industrial equipment, giving scale and more intelligence to these solutions. Some factors are:

 

  • Reduced costs for sensors

The technology used in the production of silicon chips is allowing modules and sensors to be manufactured at scale with reduced costs, meaning the cost of IIoT sensors to has fallen exponentially over the past 15 years. The following chart shows the price drop from 2004 to 2020 for industrial IoT sensors (IIoT).

 

 

Average cost of IIoT sensors from 2004 to 2020

Statista – https://www.statista.com/statistics/682846/vr-tethered-hmd-average-selling-price/

 

  • Hardware technological advancement

Before, in order to develop hardware systems, developers had to design circuits using several chips with different functions and deal with expensive, complex manufacturing analog filter circuits.

Currently, with the advancement of semiconductor manufacturing technology, manufacturers can increase the density of transistors in electronic chips and follow the trend of modularizing solutions, combining sensor, CPU and memory in a single device, simplifying development of solutions through digital and simpler interface circuits.

In addition, in recent years, there has been a great evolution in MEMS (Micro-Electro-Mechanical Systems) sensors. MEMS accelerometers, which a few years ago had a maximum bandwidth of 1,000 Hz, are now available in 6.3 kHz or higher with relatively low cost and extremely low Signal-to-Noise Ratio (SNR), suitable for precise industrial applications. This technology can replace traditional and more expensive piezoelectric sensors in many predictive maintenance applications.

Finally, the advancement of wireless technologies — such as wireless HART, ISA100, Zigbee and more recent technologies such as Bluetooth 5 Low Energy and LoRa — allows manufacture of low-energy devices, with battery life in the range of 10 to 15 years (depending on the application). This characteristic is fundamental in industrial sensors, aiming to reduce the maintenance cost of the monitoring system itself.

 

  • Scalable costs in cloud systems

It is no longer necessary for industrial companies to keep all software systems on their own servers (on premises). Currently, it is possible to count on cloud solution providers with numerous architectural possibilities, being applicable from pilot projects and Proof of Concepts (PoC) to large deployments with hundreds of thousands of devices.

The extremely flexible cost of servers allows industries to verify the real business value of very low cost solutions. It is possible to hire servers for pilot projects, in the cloud, with a high level of security with low costs such as ten Brazilian Reais per month.

 

  • Open-source solutions in Artificial Intelligence

Artificial Intelligence applications can developed in a much shorter time and with greater accuracy, compared to a few years ago. Open-source programming languages such as Python and the emergence of frameworks and libraries created by the developer community, from data analytics, machine learning and even artificial intelligence and computer vision, have popularized analytics and AI applications and led to the emergence of greater reach and interest of people and in qualification in this topic.

 

All these advances decrease costs, especially in the early stages of implementation, which, previously, could have been insurmountable impediments to various businesses. With the popularization and facilitation of access to the necessary technologies, predictive maintenance has become an increasingly accessible strategy for different industries.

 

Challenges

Although technological evolution, cost reduction and flexibility of initial costs are important factors to facilitate the implementation of pilot projects, there are some challenges in the implementation of predictive maintenance systems in the Brazilian industrial scenario. Some of the main challenges to be overcome are:

  • Legacy systems: the Brazilian industry in general has factories over 20 years old, in some cases older than 35. Predictive maintenance solutions should adapt to existing machines and not require large and expensive retrofits.
  • Choice of suppliers: in recent years many robust IIoT sensing solutions have emerged applied to predictive maintenance. However, choosing a solution that provides technological reliability, robustness and adequate support can require a lot of effort and execution of several pilot projects for partner verification.
  • Centralization, creation of data lakes, customization and cyber security: this is perhaps the greatest difficulty of industries seeking to carry out predictive maintenance projects. Integrating solutions with different technologies, working on data enrichment and data lake building, developing AI algorithms using cloud solutions — using advanced cybersecurity features — is a challenge for industries.

In the current scenario of most industries, there are numerous pilots and projects with a high technological degree, which require large allocation of the IT and AT team, generating internal competition for qualified resources.

 

For companies to be able to evolve in technology programs in predictive maintenance, an excellent solution is the partnership with innovation and technology centers specialized in Industry 4.0 and with technological capacity to meet different customization demands.

Industrial companies have different requirements regarding architecture, development tools and information security. It is critical, then, that industries rely on technology partners to assist them in generating value through the application of technologies.

 

Venturus is an Institute of Science and Technology, located in Campinas (Brazil), which has a robust structure for the development of projects in technology. We operate from the areas of IIoT, custom software development, Data Science, artificial intelligence, computer vision, cyber security and cloud architectures.

To learn more, contact us and chat directly with our industrial solutions experts.

 

Veja mais

Entre em contato

This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.
Campinas / SP - Brasil

Estrada Giuseppina Vianelli di Napolli, nº 1.185
Condomínio GlobalTech Campinas
Polo II de Alta Tecnologia
CEP 13086-530 – Campinas – SP
+55 (19) 3755-8600

+55 (19) 3755-8600
contato@venturus.org.br