Hewlett Packard's inkjet printing organization consists of multiple design centers, manufacturers and suppliers across the world. From a strategic standpoint, that made sense, but along the way development engineers began designing products with a diverse set of eyes, creating a wealth of unique designs and selection of parts. That translated into a supply chain organization that was burdened with procuring a large component mix and providing manufacturing with a diverse set of designs to support.
Higher costs resulted because designers lacked visibility into component costs and product usage, not to mention existing solutions to reuse. Inevitably, the result was proliferation in components, greater complexity, costs weren't optimized and opportunities were missed.
HP's response was to commission a Design for Supply Chain (DfSC) team, chartered to drive component and design commonality. For the team to be successful, it was critical to have data-driven decisions. After all, data was needed to convince a strong and experienced team of product development engineers that recommendations were in the best interest of the organization.
However, what started as a tool to help drive design commonality, quickly turned out to be much more as evidenced by the team’s ability to drive cost savings. In 2010, the HP Inkjet DfSC team was re-purposed from a post-manufacturing cost reduction group to a team of engineers that would engage early with R&D and have influence in the product design lifecycle. There was influence and cost savings in abundance: some $150m to date, the DfSC team says.
The team developed what became known as the RedSky Database (RSDB), a tool to provide the level of insight needed to drive commonality and optimize costs. It is described as a merging of a product’s bill of material, component costs, part attributes, forecast volume, and product attributes.
Steps to Success
The RedSky integration process involves three distinct steps:
Step 1: The production bill of materials (BOM) is imported from the product manufacturers identifying the following key elements: products, parts quantities and prices, and assembly structure information. Generally, this is 100,000 lines of data. While it’s possible to find the price of a given part, one would need to know exactly where and how to search. Furthermore, it’s difficult to find if there are similar parts that might be a better choice based on price and usage.
Step 2: To provide the next level of insight, different attributes are assigned to parts. This allows for logical organization. Examples of the different attributes are commodity type, such as screw or cable; and design characteristics, such as glass thickness or inductor value.
Step 3: In the final step, each product is assigned attributes, such as volume forecast, manufacturer and product type. This produces a data set consisting of 100,000 rows and 60 fields of data, amounting to 6,000,000 pieces of information that can now be analyzed for different scenarios.
Influencing Product Design
Now that the usage of parts in production is well understood, the next challenge is to make an impact on product development. Driving commonality means removing parts that are high in cost, low in usage and steering towards others that are optimal from a supply chain standpoint. There is no better time to switch to preferred parts than during the product development cycle. Thus, RedSky was enabled to serve R&D by making data accessible and secondly, by incorporating an algorithm to see how well the product aligns with commonality.
Having insight into the type of parts being used for a given commodity, there is now an opportunity to optimize. Component cost, purchase quantity, product usage and part design attributes are all key pieces of information that can be fairly compared with one another.
Screws used for securing into plastic illustrate the process. At one point in time, HP had as many as 28 different screws that had a thread diameter of 3mm. While there are certain properties that are diverse enough to be considered critical, there are other properties that are close enough to warrant debate. There were four screws with lengths of 9, 9.1, 9.86 and 10mm along with head diameters of 5.6, 5.39, 5.6, and 4.85mm, respectively. While one can argue that these property differences may have been critical for their design application, declaring one of these screws as preferred would be met with success for future designs. As it turned out, the DfSC team was able to identify 10 preferred screws, providing a good mix of color, length and head diameter.
The Parts Catalog was rolled out in two phases. The first focused on a select number of commodities. The format that allowed for easy searching was then tested with the design community. Once fully tested and vetted by the R&D community, it was expanded to the other key commodities.
In addition to the Parts Catalog, a Part Lookup utility was provided in the form of a graphical user interface. This provides basic queries without the need to navigate an overwhelming amount of data. Its purpose is to allow a designer to acquire business intelligence for a part that can be searched by part number, description or commodity type. Price, purchase quantity, product usage, and manufacturer usage are then presented back to the user.
Lastly, the BOM Analyzer is a tool embedded in RedSky that evaluates an itemized BOM in its entirety for commonality and preferred parts performance. Once a BOM is loaded, it makes the determination whether a part is new or pre-existing via calculation. Furthermore, it analyzes parts by commodity to determine if designers are using preferred parts or non-preferred parts.
Behind the calculations, lies a sophisticated algorithm that allows RedSky to make recommendations on whether a preferred part should be used in the event a non-preferred part is in use. For instance, if a non-preferred part is being used on a legacy assembly, it’s in the best interest of the business to not make any changes. On the other hand, if there is non-preferred part that is being used for a new design, its preferred part equivalent is identified along with other information to justify its use, such as savings, usage, etc.
What once started off as a tool to provide insight for component commonality, proved to be much more fruitful in the areas of cost optimization, price benchmarking and general analytics.
Now that design attributes and pricing for similar components are easily accessible, patterns in cost can now be seen. For example, as a result of design proliferation, the team had two A4-sized scanner glass parts being used in production, one 2mm shorter in length than the other. What was odd was that the glass with one percent less material was 10 percent higher in cost. Challenged, the supplier ultimately dropped the price by 14 percent. In addition, the team also managed to design the higher material glass out of production and settled on the lower-cost preferred glass.
With multiple manufacturers for products, pricing can be compared for the same components. Whereas two manufacturers might be charging $1.00, another might be charging $1.10. Since you can also associate these components with purchase quantity, you have the right level of insight and focus to determine if these are opportunities to investigate. Was there something unique about the supply base or were you simply not receiving competitive pricing?
The underlying ability of RedSky is the power to perform different levels of analytics which are correlated to the attributes loaded into the system. These analytics have been used to drive commonality and look for opportunities for driving down cost.
Enjoy curated articles directly to your inbox.