Featured Product
This Week in Quality Digest Live
Management Features
Gleb Tsipursky
Three reasons companies wait too long to make obvious decisions
Anton Ovchinnikov
The push-pull effects of customization
Maria Guadalupe
A new study suggests that investing in soft skills will result in higher individual—and national—productivity
Christopher Dancy
Not binding, not enforceable, but a step in the right direction
Jake Mazulewicz
There isn’t one magic bullet, but rather a ‘consolidation of subtleties’

More Features

Management News
Connects people and processes across functional silos with a digital thread for innovation
Better manufacturing processes require three main strategies
Technical vs. natural language processing
Recognized as best-in-class industry technology by Printing United Alliance
It’s unethical for companies to use test tasks as free labor
Numerous new USB3 cameras added to product roster
Combines quality and process data from software and devices in holistic production-monitoring view

More News

Andy Henderson

Management

A Perspective of the Manufacturing Future,
Part 4

Product quality

Published: Thursday, January 26, 2017 - 12:01

This is the fourth part in a series about my perspective of what the future has in store for various aspects of manufacturing. I approached each aspect by imagining what is possible using what we know to be technically possible today. In part one I covered cutting tools for machining, in part two production management, and in part three inventory management.

In this article, I’m going to describe a vision of a future for product quality. In this future state:
• A product’s manufacturability will be known during the design stage.
• As-manufactured product data will be shared between manufacturing operations to reduce nonconformances.
• Manufacturing process variability will be monitored and controlled to minimize lost productivity.
• As-manufactured product data will be utilized to optimize product performance for customers.

The current state of product quality

Today, data exist in silos. In many cases, the useful information is in the heads of a few key individuals who may not know or communicate with each other. Often, product designers have had limited experience in manufacturing. Their primary concern is product performance, and they have no good way of assessing manufacturability or manufacturing costs.

Within manufacturing, some amount of process variability and drift is always present, but when an operation is developed, it is assumed that the incoming parts are in the nominal state because, typically, there is no sharing of as-manufactured product data from one operation to the next.

Additionally, the control of the process is manual. A person will review the as-manufactured product data and decide if an adjustment is needed as well as how to make the adjustment based on experience. This person may make a decision using limited data from the history of the process, he may make an error in the decision process, and he may make a mistake when performing the adjustment.

Typically, a part has a few critical specifications for the product that ultimately goes to the customer. However, quality checks are performed throughout the manufacturing process in order to catch nonconformances at the source and before extra processing costs are incurred on a part that is scrapped. Specifications are then created to control the quality of intermediate steps, and quality checks are performed. Those specifications have an acceptable range associated with them. The stack-up of the variability from each process step can cause the end-product to be out of tolerance, even though the process steps are acceptable. Or, the specifications on the intermediate steps are conservative, ultimately leading to higher production costs. Each process step is developed assuming that a nominal part was created at the prior step.

After the product is manufactured and the customer is using it, there is a limited amount of data that returns to the producer regarding how products are used and, if they fail, the root cause of the failure.

The future of manufacturability

In the future, as designers are creating 3D models of products in CAD systems, they will be presented with information regarding the manufacturability of the features and products being designed. The quality data from the manufacturing operations inside the factories will be correlated with product materials, feature shape, and tolerances.

When similar materials are used, features created, or tolerances associated with dimensions in the design space, notifications will be presented to the designers to inform them of the capability and costs of existing processes to create similar features on existing products. This will allow designers to make better-informed decisions regarding whether existing manufacturing processes are able to produce the part or features as designed. Coupled with an assessment of customer value for the features and specifications, the designers will be able to make an informed cost-benefit assessment.

The future of manufacturing quality

In the future, as product moves through the manufacturing process, data are measured after each operation and will be used in subsequent operations to improve the product’s end quality. Typically, a part has a few critical specifications for the product that ultimately goes to the customer.

However, quality checks are performed throughout the manufacturing process in order to catch nonconformances at the source and before extra processing costs are incurred on a part that is scrapped. Specifications are then created to control the quality of intermediate steps, and quality checks are performed. Those specifications have an acceptable range associated with them.

The stack-up of the variability from each process step can cause the end-product to be out of tolerance, even though the process steps are acceptable. Or, the specifications on the intermediate steps are conservative, ultimately leading to higher production costs. Each process step is developed assuming that a nominal part was created at the prior step.

The measurements from prior processing steps will be used to adjust downstream processing steps to ensure that each step is aligned with producing products that meet the critical requirements of the product that goes to the customer. This will allow more flexibility in upstream processes because the in-process tolerances will not need to be as conservative.

If customer requirements drive more stringent tolerances, this capability may enable existing equipment to meet the more stringent requirements. This could apply beyond dimensional measurements.

For example, material composition or hardness, measured at the foundry, could be used in subsequent machining operations to optimize speeds and feeds on a per-part basis. As an individual operation output begins to drift or shift, the process control system will automatically adjust the operation to ensure that it stays in control.

Systems exist for this today, but they rely on simple rules for making adjustments. For example, if a dimension is trending high or low, then an adjustment will be made to the machine to compensate. However, in relation to complex parts, dimensions have multiple relationships to other dimensions, and adjusting one may have a negative effect on another.

Future systems will utilize artificial intelligence to recognize the relationships between dimensions and make appropriate adjustments to processing equipment.

The future of life-cycle quality

In the future, the customer will monitor products while in use, and the data will be used throughout the entire business for improving design, production, and end use.

Monitoring the quality of a product’s performance will have benefits for the customer with regards to predictions associated with product malfunctions and product downtime. This will lead to better scheduling and parts inventory management for the customer.

Other benefits will come from a data-based understanding of how the customer utilizes the product. This knowledge will drive further product development, which is more focused on customer uses. The data from real usage and any failures will be analyzed, and correlations with manufacturing quality data will be analyzed to alert factories of issues.

In this article I’ve explored a few possibilities of how quality data will be used in the future. Quality data will permeate throughout the product’s life cycle. It will enable design departments to create more producible parts, manufacturing costs will decrease, and customer data will complete the loop for improved future products.

First published on the GE Digital blog.

Discuss

About The Author

Andy Henderson’s picture

Andy Henderson

As an industry analyst at GE Digital, Andy Henderson leverages his experience from his time as an advanced manufacturing engineer within GE Power and his research during his doctoral program to promote a vision for the future of heavy industry/discrete manufacturing and drive strategy for achieving that vision.