How Prescriptive and Predictive maintenance has moved from laboratory to the field
Content
2. Evolution of Maintenance Strategy
Traditional Maintenance Strategies
Condition based Maintenance Strategies
Predictive Maintenance Strategies
Prescriptive Maintenance Strategies
4. AssetAI: Predictive maintenance Solution
Our Approach
Predictive Maintenance Technology/process
Initial Model Generation:
Retraining Cycle:
1 Introduction
The concept of equipment maintenance is not new, it probably started with the advent of the industrial revolution. With time in various heavy asset industries, the maintenance process has evolved and improved using processes, technology etc but it has not been adopted well enough to become the frontline business strategy and key differentiator for success . For long, the maintenance strategy remained at the back end operations of the company and maintenance has been done in the same old manners. Most of the advancement happened in the major university lab and company’s testbed but it did not become the part of scaled business operations due to various factors. The main cause remains the lack of evidence of value generation, viability of new technologies, implementation challenges and people’s mindset.
The aviation industry has been among the first industries to adopt maintenance as a business strategy, adopting new processes, checklists and technologies to improve airplane’s safety, uptime and overall service.
The marine industry is at the cusp of digital transformation now. Alpha Ori’s SMARTShip™ technology combined with ShipPalm (Ship ERP System) has enabled the ship and shore to digitally unite an entire vessel’s “ecosystem”, connect that to a digital cloud monitoring other ships and use the data to achieve operations efficiency, cost savings and safety of the vessel and crew.
AOT’s predictive maintenance solution “AssetAI” based on IOT and AI has opened up wide possibilities for the shipping industry to embark on a condition-based maintenance strategy to increase reliability and reduce equipment downtime.
This white paper examines the evolution of maintenance, basic concepts of predictive maintenance, the role of key technologies and what it takes to make the transformation work to realize value.
- AOT – Alpha Ori Technologies ShipPalm
- Ship management – software suite
- SMARTShip™ – An Industrial Internet of Things (IIoT) based system capable of serving multiplenumbers of vessels
- VIO – Alpha Ori’s Secure Data Exchange Platform for Maritime
- AssetAI – Alpha Ori’s Predictive Maintenance Application
- IoT – Internet of Things
2 Evolution of Maintenance Strategy
With maturing technologies like IOT, AI and big data analytics in maritime industry and utilizing it for equipment failure prediction, while improving vessel reliability and uptime, the maintenance possessed has evolved slowly over the years. The below diagram shows the evolution from traditional maintenance strategy to more proactive approach where system is able to predict and prescribe the maintenance actions.
Let’s examine various maintenance methods that has been adopted over the years
Traditional Maintenance Strategies
Maritime industry has been following the traditional methodology for years. This mainly comprises 3 major approaches.
- Calendar or Run-hour based Maintenance
The time based maintenance based on equipment run hours and calendar as recommended by manufacturer’s has been the most popular and widely adopted methodology for decades. The recommendations are provided by Manufacturers and the operating manual becomes the bible for all engineers and managers to follow. However, this is not the most efficiency or cost effective way to operate machineries and maintain reliability. Many times, unnecessary maintenance also causes maintenance induced failure that creates more urgency and affects the bottom line.
- Corrective Maintenance (CM) or Run to Failure (RF)
This maintenance strategy is adopted for assets where cost of downtime and repair is much less than that of establishing preventive maintenance measures. However this methodology can lead to catastrophic events impacting vessels’ safety and reputation.
Condition based Maintenance Strategies
Condition based maintenance is also a widely used maintenance approach in the maritime industry. Various condition monitoring technology like vibration analysis, LO quality etc is used to monitor equipment performance which was difficult to measure using engineer’s knowledge and judgment. If certain performance parameters indicate the equipment deterioration against certain thresholds.
With the advent of cheaper sensor technology, this technology will further allow more usage and adoption among the asset owners.
Predictive Maintenance Strategies
Predictive maintenance is an advanced approach to tackle equipment maintenance that utilizes the real time data using continuous monitoring technologies, machine learning techniques, and predictive analytics to identify failures much in advance and provide maintenance recommendations. The predictive maintenance solution implementation requires critical IT infrastructure and data science capabilities in place for data collection, storage and analysis to generate real value to customers.
Prescriptive Maintenance Strategies
Prescriptive maintenance is a holistic approach to asset maintenance lifecycle and one step further to predictive maintenance where systems further prescribe the maintenance recommendations based on predictive analytics but also automate the maintenance planning and execution to optimize the entire maintenance and logistic workflow. This requires the integration between IOT systems, predictive maintenance system, asset management and maintenance systems to realize the true potential of this strategy
3 Role of IOT and AI in Predictive Maintenance
The Internet of Things (IOT) platform has enabled real time data collection from various shipboard systems across a large number of ships to continuously monitor performance, benchmark them and make data driven decisions. This IOT technology combined with artificial intelligence has enabled the advancement and adoption of predictive maintenance in various industries. While IOT enables real time data collection and processing, artificial intelligence leads to anomaly detection using real time and historical data and predicts component failures. The predictive maintenance application combines these technologies to record equipment behavior, its operational status and capture any anomaly event to enable condition based maintenance and only when it is needed
The goal of a predictive maintenance application is to detect failure early and do maintenance only when it is needed. In reliability maintenance practice, most of the equipment failures follow a degradation profile that leads to equipment breakdown. This is indicated in the below P-F curve where operations and maintenance professionals have a window period to identify and perform maintenance as soon as there is an early detection of potential failure. The earlier the failure is detected, the cheaper and easier the maintenance is. This is mainly due to advancement maintenance planning, less spares consumption, reduced logistics cost and unwanted breakdown.
A P-F curve shows the health of equipment over time to identify the interval between potential failure and functional failure.
- Potential failure indicates the point at which we notice that equipment is starting to deteriorate and fail.
- Functional failure is the point at which equipment has reached its useful limit and is no longer operational.
4 AssetAI: Predictive maintenance Solution
Asset AI is a cloud based specialized predictive maintenance solution that provides real time insights, detect failures early and prescribes maintenance actions using continuous monitoring technologies, machine learning techniques, and predictive analytics
Main features include
- Real time insights by using continuous monitoring technologies (IOT)
- Advance Anomaly detection using AI/ML
- Intelligent and dynamic equipment diagnostic and prediction
- Potential failure modes identification with health score and Remaining Useful Life (RUL) prediction
AssetAI Value Proposition includes two main focus areas by using AI/ML techniques, and predictive analytics .
Our Approach
AssetAI system is an engineering solution that utilizes subject matter expertise, historical data and AI/ML in developing methodology to ensure that the model is created for predicting maintenance and early detection.
Asset AI is using machine learning (ML) method to provide advanced anomaly warning by utilizing the historical operational data collected on SMARTShip™ platform along with the past maintenance and failure data. This approach is further strengthened by including inputs from subject matter experts (SMEs) with a strong background with machineries’ operation and marine engineering. This approach and insights received from SMEs lead to faster and more accurate ML algorithms in order to achieve better diagnostics, anomaly detection and troubleshooting.
Predictive Maintenance Technology/process
The predictive maintenance technology includes below attributes or blocks of functionalities to enable AI based predictive maintenance application.
- Data Acquisition: System to allow real time and periodic data acquisition from various sensors using IOT technologies and standard communication protocols, storage and transfer to standard databases
- Data Cleansing: System to further allow data processing and cleansing to remove incorrect and unqualified data. This process requires data standardization and validation to maintain a superior data quality for analytics purposes.
- Anomaly detection: System identifies how far has the equipment deviated from normal operating behavior with the use of various AI/ML algorithms in the library.
- Health Assessment: System utilizes various performance parameters and observed anomalies to identify affected failure modes and quantifies the component health
Predictive Condition Assessment: System then combines the component health score along with the equipment maintenance history to predict the remaining useful life of the equipment in order to drive maintenance decisions.
The predictive maintenance application requires a 360 degree feedback mechanism from maintenance action in order to continuously learn and improve the accuracy of predictive maintenance.
5 What it takes for successful Implementation
It’s very clear that meaningful data in terms of number of data points and quality is critical for the successful implementation of predictive maintenance applications. If the equipment is fitted with more sensors, the coverage of failure mode is better. So, it requires significant investment in adding those to the required sensors list. However, a critical investment assessment is required in order to decide if it’s worth adding those sensors to the IOT system and implement predictive maintenance on critical equipment. In general the benefit of predictive maintenance is mainly achieved if the underlying cost of maintenance is high.
Below are some of the high level requirements in order to get the full benefits of predictive maintenance application.
- Equipment is fitted with good sensors list for better asset management
- Equipment has overcome the infancy cycle in terms of its lifecycle.
- Availability of equipment maintenance and failure history
- Ability to include additional sensors in future
- PMS integration capability in future
With the capabilities of predictive maintenance, it’s possible to monitor equipment health and its remaining useful life for vessel’s reliable, safe and cost efficient operations (i.e. reduced spare part consumption, minimized inventory levels, energy efficient operations etc.) leading to a more sustainable future for shipping. It will further lead to a tangible impact on shipping business operations by improving vessel reliability, reducing equipment downtime and avoiding any safety and regulatory noncompliance.”
About the author
Uttam Kumar is a Director, Product Management at Alpha Ori Technologies (AOT) and leads SMARTShip™ product management and predictive maintenance applications. Contact at uttam@alphaori.sg