How does data warehousing apply to machine monitoring information? The automobile company I work for collects machine states and events real time and stores this information for later analysis for productivity improvements. After a year's worth of data is collected, the data is purged. What value can data warehousing provide?
Chuck Kelley's Answer: Keeping a long-term view of the states and events of your machines provides the ability for lots of different types of analysis. Remember that once the data is purged, it is lost. You might use a longer-term view for analysis of productivity improvements and/or quality analysis. For example, if you found that a change to widget A was needed and you did it (along with forecasting the expected results), then after collecting the new widget A data you can compare how much you have gained. Then you may make other improvements. You might be able to understand the effects of changes over the rest of the machine. Sometimes, there is an unexpected loss in one machine due to a productivity performance of another.
Tom Haughey's Answer: Data warehousing is an environment for gathering data over time (even long periods of time), storing that data in a reusable database and delivering it as information to business and IT users. A data warehouse has three major processes: gather, store and deliver. It is used for analyzing that data and reporting against it. While data warehouses store data over time, some warehouses (soft drink distributors are an example) might roll off data after two to three years. Data warehouses should be used to report on what happened (stage one), analyze why did it happen (stage two) and predict what will happen (stage three).
I suspect that machine state and events data could be used to determine the productivity and quality of people, processes, workstations and products. However, even decades of machine data alone is only of so much value. I propose that this data needs to be enriched with other data to see its full potential. It could be enriched by adding data from other business processes, adding more dimensional data (perhpas about employees, plants, processes, products) and even perhaps buying data from the outside.
I would talk to other mangers in your business area and related business areas, such as human resources, benefits, disability, manufacturing, product quality. Then ask them, "What are the top-ten questions you need answered?" [See my answer this month on "Risk."] Get them to be as tangible as possible in their answers - 10 different answers, from each manager if possible. Then ask, "What is this worth to you?" Add it up. [Again, see also my answer this month on "Risk."] Little by little, this will allow you to begin to form a picture, from which a strategy can eventually be derived. The thing to do is begin to ask management what answers they need and what value those answers are to them.
Consider this example. How much disability do you pay? Surely and unfortunately, people get injured around machines. They have to spend time with doctors, hospitals, occupational therapists, etc. This costs a bundle. Suppose you could reduce disability by one, five or 10 percent. What would that be worth to you? How would you do that? By analyzing current disability incidents and patterns, then finding employees or situations that have those characteristics (and are thereby likely to end in disability), and seek to address them in advance. I wouldn't be surprised if in large auto manufacturer you could pay for the warehouse in one year by reducing disability payments one percent.
In addition, I suggest focusing on other processes that relate to your machine data. Once enriched it could be used for many other purposes, such as HR. It could be used to determine qualities of good employees, the characteristics of good managers, disability understanding and avoidance (as we saw). The main closely related business areas I would look at would be productivity, HR, product quality, vendor quality, manufacturing and logistics.
Register or login for access to this item and much more
All Information Management content is archived after seven days.
Community members receive:
- All recent and archived articles
- Conference offers and updates
- A full menu of enewsletter options
- Web seminars, white papers, ebooks
Already have an account? Log In
Don't have an account? Register for Free Unlimited Access