Cut costs with predictive maintenance.

Implementing a predictive maintenance program can yield cost savings throughout the plant.

In many foundries preventive maintenance is a well-established practice. Some of these facilities are now moving on to predictive maintenance. Predictive maintenance is defined as "the collection of data and use of graphic representations of measured parameters to indicate equipment condition." This information then is used to restore the equipment to good operating condition.

Vibration Analysis

One of the several predictive maintenance tools in general use today is vibration analysis for equipment. This information is down-loaded into a PC for analysis. The information can be viewed in tabular or graphic form. (Graphic analysis often is chosen because it indicates trends in pictorial form.) The system uses trend analysis as a basis for action.

Some computerized maintenance management systems (CMMS) have now made their software compatible with vibration analysis software so that when predetermined parameters are reached, the CMMS automatically issues a work order for repairs.

Lubricant Analysis

Another predictive maintenance tool gaining in usage involves the analysis of lubricant samples taken from gear reducers, hydraulic reservoirs and other critical areas. This procedure is used to monitor wear and lubricant quality to detect problems caused by adhesion, friction, contamination and corrosion. Statistical analysis of this data will allow corrective action to be taken (i.e., changing or filtering the lubricant and replacing defective components) before major damage occurs.

Thermography 

Thermography also is used as a predictive maintenance tool. This involves the use of an infrared detection unit to scan electrical systems. The detector will reveal any "hot spots" indicating a bad electrical connection. Since many bad connections are inside buss ducts, conduit or other enclosures, they are not apparent until they fail and downtime results. By detecting these hot spots, the necessary repairs can be made before failure occurs.

New Technique

A new predictive maintenance technique (SEMCO--Statistical Equipment Maintenance Control) has been introduced for maintaining quality in production areas. Statistical process control (SPC) and statistical quality control (SQC) are now well established as tools for maintaining quality in the production areas. However, their use in predictive maintenance is just beginning.

SPC involves the statistical analysis of measurements to determine the process variations over a period of time. In production, this involves the sampling and measuring of dimensional variances in parts. In the maintenance function, the same principles can be applied to determine variances in the equipment. Since anything that can be measured accurately can be analyzed statistically, this becomes the basis for using predictive maintenance techniques throughout the plant.

To use SEMCO for predictive maintenance, one must first establish parameters for each measurement, equipment operating characteristic or set of measurements that will affect the quality of the product. This may be an actual dimension, parallelism between surfaces, a machine level, hydraulic pressure, a current reading, etc. The measurements for new equipment are easily determined from the manufacturer's specifications. In older equipment, it may be necessary to adjust the equipment to as "near new" condition as possible, then use those measurements as the starting point. It will be necessary to qualify at least 10 measurements of the "established" parameter to test accurately.

As the equipment is operated, the measurements are repeated over a period of time. When the machine wear has caused deterioration to the point that the equipment is beginning to make scrap, then necessary repairs should be made. Over an extended period of time, this process will establish control limits that can be shown graphically on a control chart. There are several good computer programs available for statistical control. The plotting can be done manually, but can be time consuming.

Applications

A SEMCO chart plotted from actual readings taken from a molding machine is shown in Fig. 1. At Point 3 on the chart, the deviation of the dimension was out of the established parameter (above the upper control limit (UCL) for the dimension). At this point, corrective action was taken, and the next reading was well within the allowable dimensional range.

At Point 18, an adjustment was made, no parts were replaced, and again, the measurements were back within range. Although Point 18 was not out of the specified limits, action was taken to prevent it from exceeding the limits.

As in other types of predictive maintenance systems, the readings give a trend analysis showing well in advance that some future action will be needed. In this case, the readings were taken weekly. The developing trend indicates the machine condition long before the machine condition long before the machine starts to make scrap parts so that corrective action can be taken before scrap is produced.

SEMCO charts used to monitor molding flask pin and bushing wear are shown in Fig. 2. An automatic recording micrometer measured the pins and bushings and collected the data. Both charts indicate that excessive wear occurs after 180,000 iterations. To prevent casting defects due to mold shift and core crush, it was found that the pins and bushings should be replaced at or before this point is reached.

Benefits

Predictive maintenance techniques benefit production, as well as maintenance. In production, one benefit is the realization that equipment is being restored to and maintained at nearly new condition. This is certain to reduce scrap. Well-adjusted and maintained equipment will produce parts within the capability of the machine. This alone may more than justify the time it takes to implement the predictive maintenance program.

Another gain for production is reduced unscheduled downtime. Since predictive maintenance points out trends that indicate the beginning of a problem, repairs and adjustments can be made before they cause downtime. Lost production time often is greater than the cost of the repairs. There are even more benefits that directly affect the ability of the maintenance department to perform their duties.

An advantage of the SEMCO program is that it helps establish benchmark measurements of selected parameters. These measurements then indicate when worn parts must be replaced or adjustments made. Having a record of the machine's capability means that the adjustment/replacement function is done when, and only when, it is required. This eliminates the guesswork that has been prevalent without predictive maintenance.

The next benefit to be realized is the establishment of the machine capability. This is necessary information for both maintenance and production. The knowledge of the true capability of the equipment to produce parts within certain tolerances will forestall unrealistic demands to produce parts with tolerances tighter than the machine's capability. Not only will this reduce maintenance time, but it will force the assignment of close-tolerance work to machines that are capable of doing the job.

Another benefit for maintenance is the elimination of wasted time in troubleshooting equipment. Scrap is often the result of several causes acting on the machine simultaneously. Any machine functions that are checked using predictive maintenance techniques are eliminated from the causal factors. Thus, maintenance personnel will not waste time checking items known to be in good operating condition. Troubleshooting will be limited to areas that are not being checked regularly.

Predictive maintenance also helps to minimize excessive parts usage and the related costs. (It is estimated that 30% of controllable costs in a manufacturing facility are maintenance costs.) Maintenance personnel have long been aware that poorly adjusted parts will wear out at an accelerated pace.

Predictive maintenance gives an early indication that an adjustment is needed. Thus, parts can be replaced before causing damage to surrounding parts. If worn parts are not adjusted or replaced before a breakdown, excessive damage can result. The cost of the resultant downtime will add further to the overall repair costs.

Implementing a predictive maintenance program becomes very cost effective and is recommended for proceeding into the total productive maintenance (TPM) concept.

Before implementing predictive maintenance techniques, it is necessary that the personnel involved receive adequate training. If no employee is able to train co-workers, it will be necessary to obtain training from an outside source. There are many books, seminars and training courses available on these techniques.

Note: The SEMCO predictive maintenance technique was presented as part of the CMI Productive Maintenance course at AFS on February 18-20, 1992. We thank R. Welker, Newnam Mfg., Inc., Kendallville, Indiana for the contributions on pin and bushing wear.

For more information on the SEMCO predictive maintenance technique circle No. 343 on the Reader Action Card.