4. Asset management and optimization novel approach
4.2 Asset management and optimization approach enhanced with predictive
4.2.3 Improving work processes
This component of a maintenance management strategy consists of documenting and tracking the maintenance work that is performed. This involves the use of a work order system to initiate, track, and record all maintenance and engineering activities, including predictive maintenance [54]. The inclusion of the predictive activities is essential to track the amount of resources being expended while performing PdM activities. If data is lost, true analysis can never be performed [13], [21].
An effective planning and scheduling are important to the predictive maintenance program for two reasons. The first is that maintenance technicians need to be scheduled for the PdM activities. The resource requirements for the PdM program need to be projected and tracked through the work order, planning and scheduling functions. The second is that the PdM program will identify work activities that need to be performed on equipment so that an unplanned failure cannot occur [46], [54].
Work processes are an important aspect of asset optimization. Substantial cost reduction even before introducing new technology can be achieved by examining current maintenance work processes and the way things are normally done [42]. Well- organized maintenance work processes comprise four phases:
• Initiation/prioritization/purging: diagnostic information must be used to evaluate potential projects and determine in what order they must be executed. Purging is useful to get sure if the new project adds value to the plant and whether it pays off [48];
• Scheduling: capable personal and necessary material must be brought together at the right time and place. “Careful scheduling can eliminate wasted effort and cut 20 to 30 percent off the time required to complete a job” [48];
• Execution: the use of asset management software can accelerate the execution of daily maintenance tasks and save time;
• Analysis: an analysis based upon accurate data collection, root cause analysis and standard reliability principles must be carried so as to determine if maintenance is a vital need or not. Afterwards, this information must be used for future decision-making.
This novel maintenance approach incorporates CMMS. Reliable information is directly received from plant assets and is scanned in real-time for alerts [48]. The on-line software is charged for delivering the data and serving as trigger to launch CMMS transactions [1], [48]. In most companies, there is sufficient work order data accumulated by the maintenance and engineering functions to require the computerization of the data flow [46]. This facilitates the collection, processing and analysis of the data, and it provides information support for the Maintenance Management Strategy.
From a Predictive Maintenance perspective, the CMMS system is normally interfaced to the predictive software system. There is usually an interface that allows triggers or alarms in the predictive software to generate work orders to make repairs that are identified by the PdM inspections [47]. The work is tracked through the CMMS system allowing data to be posted in the equipment history, where analysis such as mean time to repair (MTTR), mean time between failure (MTBF) and total cost can be performed [40], [41], [48].
When predictive maintenance technologies are integrated, users can automate the maintenance process from point of alert to completion of the maintenance work order. Using Web-based IT platform can facilitate analysis and accelerate decision-making [50]. New technologies collect, consolidate and distribute asset information to the people, whenever they are located [44].
4.2.4 Maintenance decision-making
The PdM decision on when to take equipment down for maintenance should be integrated with overall manufacturing operations decision making and not viewed as an equipment-only decision [51]. This is feasible by providing a holistic view of all the processes. The challenging goal is to have 100% confidence in the information from the tool about current and predicted performance, which will reduce the variability of equipment availability [54]. With the large amount of data requiring analysis, reliable data transfer is critical to enable intelligent failure prediction [13], [21]. Novel solutions are required to integrate equipment data and overall factory operations into valid data sets for PdM decision making. Failure models for each key component of the tool
should be standardized. Model-based-predictions must use historical and real-time data from the tool to generate an imminent failure warning [46]. While the component is running, the model should compare current performance to the result that is expected.
Once data from all the predictive maintenance technologies have been acquired and correlated, a fault may be found. If the machine is surely going to fail, it is necessary to shut it down immediately and replace the damaged part. The following tools enhance and enable to perform efficient decision-making concerning the maintenance of the assets.
• Condition monitoring and health assessment- A baseline for equipment performance must be established so as to enable predictive maintenance solutions. Standard condition monitoring techniques are employed primarily in process monitoring [51]. Furthermore, “condition monitoring algorithms” that include equipment signals integrated with statistical process control must be developed in order to deliver health assessment metrics for the assets;
• Health assessment and prognostics- Prognostics is one of the key challenges of predictive maintenance technologies. The mathematical algorithms aimed to detect impending failures are so complicated that they need AI learning solutions. “Due to the complexity and uniqueness of each tool family, prognostic algorithms may need to be developed with each equipment supplier” [51]; • Data integration and automation architectures- PdM final decision must be
based upon the integration of multiple data sources. This integration process constitutes a major challenge for manufacturing plants. The sheer volume of data can be overwhelming, particularly if integrated data sets are required [50]. This fact must drive innovative approaches to equipment data collection and treatment. It must not be forgotten that delivering a PdM solution flexible enough to works across each wafer size technology presents a major challenge, due to the wide variety of existing data sources, quality of data, and automation architectures [49], [51].