Reading the bottom of a coffee cup — Production Log Analysis

Did your production dashboard tell you this? 50% of the orders have greater than 90% or higher shortages, while machines are not exceeding 45% utilization and 50% of the staff are about 10% utilized.

In the past, time and motion studies were expensive, but not today with all the IT systems that exist. Most IT systems today maintain a copious set of logs. Technologies exist today that enable us to even reverse engineer the process flow from these logs. Saving specialists time to explain all possible scenarios that occur.

It may seem that much information will be required, on the contrary, little is required for such a study. This methodology can be applied to analyze almost any of the existing processes in an organization. Following is a case from a manufacturing company’s machine shop.

So what is the data required?

Here is a structure

Suggested data format for analysis

Below is a sample for an order / Case ID

Life-cycle of a single order/ Case ID

Observations and Conclusions

Let us look at what can be determined by merely analyzing the logs — to this point there is no discussion with any client representative.

• The logs start on Jan/2/2012 till March/31/2012. There are 65 working days in this period, considering 5 operating days per week and no holidays.
• Logs tell us that the machine shop is operational 24 hrs while the personnel are operating in 3 shifts of 8 hours.

The machine shop “possible” layout

Possible Machine shop layout based on data made available

The count of machines and their uses is determined from the logs. 2 machines are under maintenance and their functions are unknown.

The basis of layout and machine placement is from the observed process flow.

Utilization — Machines and Workers

• The machine shop has a total of 26 machines.
• 2 have a high usage exceeding 90%. That is, they operated for 90% of the available time.
• While 15 of them are either under maintenance/ their utilization is lower than 10% of the available time.
• Observation 1: why is the capacity utilization low of machines?
• Observation 2: why is order fulfillment low despite low utilization?
• Observation 3: why is there an imbalance of turning and milling machinery — is it that machines with low utilization are faulty? Or specialized for a specific task?

• There are also a total of 44 workers in the machine shop
• 11 of the workers have a high utilization of 95% or greater.
• There are 21 workers whose utilization is less than 10% of the available time.
• Observation 1: There is a lack of task specialization; any worker operates any machine
• Observation 2: 75% of the workers have spare capacity. Disclaimer: These may be temporary workers or working on the main production floor is unknown. This should be discussed with the client.

Order Fulfillment view

• The machine shop operates to fulfill the ordered products and quantities. Here is a view of the fulfillment rate:

• 39% of the orders have supplied quantity, within 95% of the quantity ordered.
• Almost 49% orders have the supplied quantity, lower than 90% of the quantity ordered.
• 51 orders have 0 quantities created.
• Observation: why are there so many orders that consume time but have no output?

Listing products that took 80% of total time is as follows:

Summarized information by Product

• Cable Head and Ballnut — took ~45% of the time but only averaged about 50% of ordered quantity.
• Disregarding single orders, Spur Gear is the only product that has multiple orders and a high fulfillment ratio (supplied/ordered).

As-Is Process flow

The process view constructed from the logs looks something like this.

Animated Order flow —over 3 months

It depicts the various activities and the flows. The numbers on the flow lines are the cases/ orders. A total of 225 orders start from the left and finish on the right. The darker the shade of blue, the more frequent is the activity. Turning and Milling QC, Lapping and Flat Grinding are the top set of activities that happen. The count of variant paths is reduced to create a more concise process flow. The flow, in red with numbers, indicates the cases that have taken that path. Red flows on an activity indicate a count of cases that went through repeats of this activity.

A more telling picture is the one below that indicates the time about those activities. The mean time required from start to finish is 20 days. There are cases that have taken much less or more and are considered outliers.

Animated — Flow of orders and the time they spend in each activity

The darker the red, the longer is the time taken by that activity to complete. Stand out and require a client discussion, if these are bottlenecks, are: Turning and Milling, Turning and in that order of red. There are 2 more animated files (enclosed separately) that show the flow of orders. Each yellow bubble is a case, pile up of these indicates the slowing down of the process and can be evaluated for improvement/automation.

Variants —different sequence/ steps performed to make a product

Let us look to the table on the left:

Focusing on Spur Gear, we have 26 orders and there are 26 different (#Variants) activity flows adopted in completing the orders.

Similar behavior is observed for Ballnut and Cable Head. 58 orders, 56 variants and 50 orders with 49 variants.

Conclusion: Lack a standard operating procedure for a product.

In summary

1. We understand in depth a process and its variants by analyzing application logs.
2. Obtain insight — without having to speak with specialists or searching for “updated” documentation.
3. Identify existing/potential bottlenecks for improvements.
4. The ability to benchmark and quantitatively evaluate improvements/corrections performed.