Smart Machine Pilot 2 Project (SMP2P)

Background/Problem

Today, valuable information lies buried inside individual machine controls on the shop floor. Extracting information from these isolated islands is difficult and time consuming. Lean staffed, multi-shift operations make it difficult for operating personnel to maintain an accurate situational awareness from one shift to the next. Maintenance personnel receive scant advance warning and are called to diagnose faults based on a filtered verbal explanation of the problem. Loss of older experienced personnel causes a loss of institutional knowledge. Upstream processes (design) often don’t receive feedback from downstream processes (manufacturing). The cost/benefit ratio to attempt to gather, compile, and analyze this buried information using conventional means is too high.

As a result, operating, support, and management personnel are left to make decisions based on incomplete and subjective information. Problems and inefficiencies remain hidden. Opportunities for “continuous improvement” are not realized because the supporting facts are not known.

• What is the true short-term and long-term performance history of this machine? Is it changing?
• What is the cycle time and cost to make this part? Has it changed? Is the standard time correct?
• Has the part process consistency changed? Is the machine being over (or under) stressed?
• What cutting tools am I using on which machines? Am I getting the best usage in all cases?
• Is that machine starting to show health problems? Will it be available when I need it?
• Am I getting the best utilization from this machine? When has it been idle and why?

Solution/Approach

The Smart Machine Phase II project will equip different selected machines and/or processes at additional project partners’ sites with the capability to automatically gather and report their performance in a consistent fashion. This performance data will provide the necessary means to understand trends, predict potential delays, and allow shop operations to visibly improve their equipment and/or process utilization.

Of significance, is a new single-platform monitoring system that has the flexibility to monitor both process and equipment performance, regardless whichever is being evaluated.

Operating and management personnel can now access an interactive presentation of production process data not previously available. At the same time, autonomous software processes operate behind-the-scenes to traverse the accumulated volume of this data. These processes gather, record, analyze, and present pertinent factual data. They serve to identify anomalies and report pending problems in a timely fashion to targeted individuals on a need-to-know basis. Better information enables better management to achieve the goal to make parts better, faster, and at less cost.

This project will also benefit through collaboration with the Department of Commerce’s National Institute of Standards and Technology (NIST). NIST will contribute data it has collected in its Manufacturing Engineering lab. The National Science Foundation funded Intelligent Maintenance Systems Center will use their expertise to analyze data from NIST and Caterpillar with the intent of developing prognostics algorithms. This short Phase I effort is expected to lead to a more comprehensive Phase II proposal where the lessons learned here will be applied to more sophisticated applications such as prognostics.

Anticipated Phase II Benefits

Process Variation, Efficiency, and Health
For the RRAD manual rubber compression molding process, cycle time variations of as much as 25% were noted. Once it was known that cycle time was monitored, variation dropped immediately. In Phase II the system will be used to examine details of the process, such as internal mold temperature over time, with the objective of defining in detail the bounds of acceptable process parameters. The obvious next step would be to pilot automation controls on one or more machines, but until the BRAC process is complete work beyond Phase II at RRAD is questionable.

Access to archives of actual data collected autonomously from production equipment can yield significant benefit in improvement to equipment utilization. One mid-sized machine tool user applied such data to improve utilization from 40% to 80%. The keys to this result were:

1. Data was collected autonomously so there was no human interpretation
2. Data was archived and presented in time-based graphs so trends were easy to spot visually
3. Data was collected on a periodic basis with small time increments between polling events so transient events were also recorded
4. Reports were easily accessible via Intranet for easy management tracking.

Of course 100% utilization improvement isn’t feasible for all installations, but it is safe to predict some level of improvement for all installations. In the depot environment improvement in equipment utilization translates to greater throughput and that in turn translates to improvement in depot maintenance cycle time.

Work at the Caterpillar diesel engine transfer line is focused entirely on equipment health and integration with maintenance management systems. The objective is to improve equipment up time since by the nature of transfer lines a problem stopping one station halts the entire line. The single section instrumented in the abbreviated Phase I did not yield sufficient data to predict the magnitude of future improvement, but improvement is certain.

The bottom line is that access to accurate data, logged and trended over time, provides key insight into root cause problems affecting equipment utilization has always resulted in improvement. The magnitude of the improvement will vary, but any improvement lowers costs and reduces cycle time. The installation at OO-ALC targets the B-2. With only 21 aircraft in the total B-2 Fleet, any reduction in cycle time will reduce part delivery time and be reflected in B-2 Fleet readiness.

Integration with Maintenance Management Systems
The work at the Caterpillar diesel engine transfer line represents the sole site in which the data capture system will be integrated with the plant maintenance management system. The result will be that control system alerts can autonomously initiate maintenance work requests for those alerts that require intervention. Maintenance managers and technicians will also have access to archived data that can provide a permanent, easy-to-view record of all alerts and faults. Automating maintenance requests and providing easy access to machine performance data by maintenance personnel can also increase equipment up time.

Predictive Algorithms
No work was performed in Phase I on algorithms to detect fault events before happening allowing corrective actions to minimize major shut-downs of equipment. Technologies exist that can learn over time the sensor signatures differentiating good equipment behavior from degrading behaviors. The objective is to recognize impending faults and initiate that corrective action. Phase II databases of equipment data will be much more extensive, enabling early work on predictive functions.

The Bottom Line for Benefits
Quantifying projected benefits for this project is difficult for two reasons:

1. This type of data collection and presentation system is relatively new so detailed data from other installations is difficult to find
2. Hard data on depot equipment utilization and correlation of that data to depot program maintenance (DPM) cycle times is even more difficult to find.

If one quarter of the total DPM time is spent actually in process on depot equipment and if that equipment utilization can be improved by even as little as 10%, then it is reasonable to expect a 2.5% decrease in DPM time. If DPM for a major system such as a KC-135 is on the order of 200 days, the same improvement could shave 5 days off the total time in maintenance.

Program Manager: Tony Haynes, (734) 995-4930, tonyh@ncms.org