MBD Implementation Dos and Don’ts: Verify and Validate MBD Data

In a previous article, I introduced the model-based definition (MBD) implementation dos and don’ts. Table 1 shows a collection of several recommendations from the people, process and product standpoints based on the practical experiences of dozens of manufacturers around the globe. To learn more details, please follow the embedded links to the previous posts. In this article, let’s continue this series and look into the verification and validation of MBD data.

Table 1. The Dos and Don’ts in MBD implementation.

Key Aspects Dos Don’ts

·         Drive from the top down

·         Establish a core implementation team

·         Motivate all involved

·         Don’t hesitate (Part 1 and Part 2)

·         Don’t take too big a bite at first

·         Don’t be confined to the Design department alone

Process ·         Measure progress

·         Automate workflows

·         Update infrastructure

·         Don’t rely on 2D drawings as the master anymore

·         Don’t only send 3D PDFs to the supply chain

·         Don’t overlook MBD over the Web

Product ·         Organize and present 3D PMI clearly

·         Customize and standardize templates

·         Don’t skip critical 3D dimensions and tolerances(Part 1 and Part 2)

·         Don’t stop at graphical 3DPMI(Part 1 and Part 2)

·         Don’t exclude printouts

To put things into perspective, let’s begin with a supply-chain story. Five years ago, a vendor was asked by a client to adopt the MBD process. For a selected pilot product, the client built a new model-based design using a proprietary CAD tool and exported the design into an intermediate 3D format. The MBD data in the intermediate 3D format was sent to the vendor, including the models and integrated 3D product and manufacturing information (PMI). No 2D drawings were provided.

To fulfill the order, the vendor followed the mandate and plowed ahead. The supplier’s engineering team found a viewer of this intermediate 3D format to open the data and interpret the design. However, the team kept running into obvious mistakes in the client’s data, but they weren’t sure whether the issues were caused by the data, the viewing software or the usage of the software. Finally, several months later, the vendor raised a red flag to the client—only to discover that the files they received didn’t match the client’s original design. Now after months of time, money and work had been wasted, the finger-pointing began.

I wish this incident was isolated, but it’s not. In an MBD validation pilot conducted by the Department of Defense and the National Institute of Standards and Technology (NIST) in the United States, one of the pilot projects was the Bradley cross-drive transmission as shown in Figure 1. The research team found that 21 percent of the transmission models contained errors.

Figure 1. The Bradley cross-drive transmission models (left) and the physical product (right). (Image courtesy of Paul Huang.)

How then can we avoid these kinds of errors? In the 2D drawing–based process, there is a step to check the drawings. In the MBD process, we need a similar checking process to both verify and validate the MBD data through manual review and/or using automatic software tools. Let me explain these two steps—verification and validation.

Verification focuses on checking the cohesiveness of the model,as well as the product and manufacturing information (PMI) in the source 3D CAD design. For example, a common question during the design phase is whether a part can be manufactured on the shop floor, or what kind of manufacturing issues might arise from certain types of design. Figure 2 shows a manufacturability check by an application called DFMXpress in SOLIDWORKS.

image003Figure 2. A manufacturability check by DFMXpress.

Besides the model manufacturability checking, here is another example with PMI. If a maximum material condition (MMC) modifier  (M in a circle) is attached to a feature positional tolerance zone that does not have feature size tolerances defined, then there is no way to identify the MMC. Therefore, this modifier results in an error. SOLIDWORKS MBD can catch this issue as shown in Figure 3. There are hundreds of geometric dimensioning and tolerancing (GD&T) rules built into the software to check errors. More details can be found in a previous article, “Check Your Grammar: Verification for GD&T and MBD.”

image005Figure 3. A GD&T MMC modifier error is flagged in both the graphic area and the tree nodes in SOLIDWORKS.

In contrast, validation checks whether a derivative matches its original source. For instance, if a hole is missing or if a tolerance value is altered from a CAD model to its STP export, it should be caught by the validation step. A presentation on MBD implementation practices by Casey Gorman with Sparton shared these experiences:“Before sending data to suppliers, we validate STP derivatives against the original CAD models and use a CheckSum tool to protect them. Later when a supplier sends us their STP or models used for manufacturing, we compare their models against ours. If they changed our models and the part didn’t pass quality checks, they would have to eat this part, instead of us.”

Similarly, Gulfstream established rigorous processes to ensure data integrity. A software application (Kubotek Validation) is used to run the topology analysis by comparing model surface changes from one CAD version to an upgrade, or from the CAD file to its derivative.Furthermore, another software tool (ITI CADIQ) is deployed to compare the point changes. Besides models, these tools are also used to compare the PMI differences.

If the separate results by Kubotek Validate and ITI CADIQ both confirm zero changes, then this dataset is regarded as well preserved between software versions, or between the source data and the export. If there are any changes to the dataset, or any inconsistencies between the validation results by the two different tools, then the software upgrade can’t take place, or the data export is not trustworthy until the issues are resolved.

Now let’s circle back to the supply-chain story noted above.For clients, it’s highly recommended that models be verified and any derivatives be validated against original designs before sending them to suppliers. For suppliers, it is best to work from a client’s verified and validated data. If a vendor’s engineers have to recreate any new models, then the new creations must also be validated against the client’s data. These verification and validation requirements may even be built into a legal contract before a job is accepted to clarify the rights and responsibilities of parties.

There are capabilities in SOLIDWORKS today to help you build trust into your data, such as DFMXpress, PMI verification and Design Checker. To learn more about how SOLIDWORKS MBD can help you with your MBD implementations, please visit its product page.

About the Author


Oboe Wu is a SOLIDWORKS MBD product manager with 20 years of experience in engineering and software. He is an advocate of model-based enterprise and smart manufacturing.

Recent Articles

Related Stories

Enews Subscribe