Pumps & Systems, September 2007

Using root cause failure analysis to address equipment design and manufacturing deficiencies that increase pump reliability and reduce equipment failures.

Define the Problem

With any problem-solving technique in the chemical industry, understanding the scope of the problem is the important first step in defining what improvements are expected. This helps ensure the team will follow the correct path together, rather than vying for each person's own agenda.

For example, a pump may create trouble in a process and cause downtime for operations. The most important thing for an operator is to have a pump that does not fail, but from a maintenance standpoint that same failing pump may suffer from poor quality parts, operation beyond the design conditions, or other aspects that can contribute to the failure modes. Together, the operations and maintenance teams must agree on the scope of the problem before heading down the path to the subsequent steps.

Gather Data

In simple terms, you don't know what you don't know. More specifically, try to understand the problem and gather data and other evidence on things that are currently unknown or have not yet been proven. This typically requires lots of legwork to gather field data.

Quantitative information, such as operating conditions, combined with qualitative information from interviewing different people can help fill in the missing pieces of the puzzle. In some instances there may be one group that performs a task differently on the system being studied - and that one difference contributes to the issues being studied.

This step is a good time to confirm with the OEM exactly what the design criteria are and what the consequences are of exceeding any design limits of the equipment. Defining the acceptable operating envelope and criteria are most significant. User manuals, industry guidelines, national codes, national standards, best practice databases, and supplier technical support are all avenues to use in evaluating the equipment's original design envelope.

Failure Analysis

Using a methodical and sound failure analysis technique helps to start bringing the pieces of the puzzle together. Having a broad spectrum of people with different expertise and viewpoints facilitates quick identification of different failure modes to consider in the analysis.

Some of the data already gathered may help to support or refute the potential root causes discussed in the failure analysis meeting. Other items identified may need more data gathering or third party analysis to help validate or determine the root causes, such as advanced metallography, finite element analysis, or material property testing for tensile strength.

At the failure analysis stage, it is wise to invite the local OEM representative to attend the meetings and bring their expertise to the table. In some cases, the vendor may have already faced this same problem and have a solution ready to implement. Other times, the vendor is not aware of the problem and may need to start requesting help from the OEM's main technical support staff. Coupon samples or the entire damaged part may be requested by the OEM for further detailed analysis by their facilities.

Providing the OEM with the failed components is a good first step toward building a non-confrontational relationship with the OEM that will leverage their support and expertise. However, we advise that some portions of the parts be held or sent to an independent third party for evaluation if it becomes necessary to obtain independent confirmation of the required testing and analysis.

Involving a purchasing agent may not seem logical at this stage, but opening the communication line to the person responsible for buying the equipment being analyzed helps keep them in the loop. This communication also makes it easier during later stages in the problem resolution to place pressure on the vendors and OEMs for help in solving the technical problems.

A documented timeline of when the problem began and regular updated progress reports help the purchasing agent determine whether the OEM is being supportive and expeditious or dragging their feet to address the problem. Contractual stipulations may exist within a corporate supply agreement requiring the OEM to provide a certain level of service to maintain good standing for future business opportunities.

At the end of the failure analysis process, a working list of corrective actions items should be generated and a responsible person assigned to each one with a due date. If possible, a key contact person for the vendor or OEM and a key contact person for the end user should be designated to help keep the lines of communication streamlined.

Progress Reporting

After the initial rush of data gathering, failure analysis and meetings passes, it is easy to assign the corrective actions and walk away, thinking that everything will take care of itself. Agreeing on regular dates to have progress reports on the action items is crucial in the early stages of the process. This progress report helps clarify any stumbling blocks that may appear which were not considered in the initial meetings. For example, a certain test the OEM was prepared to perform may not be feasible, or a larger sample may be required.

In other situations, the end user may change campaigns unexpectedly and be unable to gather more field data until a later date when the campaign ceases. Good two-way communication between the end user and the OEM is a key for continued progress at this step.

Regular updates also help the purchasing agent monitor the timeline for the required support from the OEM and allow them to step in for any necessary assistance. The progress reporting can be logged using any method and software which streamlines data exchange between the parties. A Microsoft Excel spreadsheet has been successfully used, but other forms are just as acceptable. The main point is to keep a logical track of the information in one consolidated document.

Problem Solution

Once all the data is collected for the problem equipment and the failure modes analyzed and discussed, the team should have a better understanding of the problem scope and be able to develop a final list of recommendations or corrective actions.

Not all of the recommendations may be financially feasible or practical for the end user or the OEM. A recommendation may involve a complete part redesign, but the OEM tooling would be too costly for this one application so the OEM may choose not to change their design. An alternative may be smaller design changes or modifications that still yield acceptable results, even though they are not the ultimate solution.

Some design modification recommendations may be readily accepted by the OEM due to an advantage that makes the part more robust for other customers and provides added value to the product. In this case, both the OEM and the end user benefit from the teamwork.

Some modifications may also have to be made by the end user to ensure that the equipment is properly installed, operated, and maintained. In some cases, making process changes that adapt to unresolved OEM design and manufacturing shortcomings may be possible - or even necessary. Procedures may need updating. Operator and mechanic training may need to be scheduled. The OEM and local equipment representatives should be requested to take part in this training so the workers can successfully obtain the maximum benefit.

Summary

A simplified work path has been outlined (see Figure 1) in order to perform an analysis on an equipment item that causes reliability problems in a manufacturing facility.

Figure 1Key steps that include requesting of support from the OEM are: 1) the initial failure analysis, 2) the evaluation of the failed components, 3) support necessary for generating corrective actions, 4) providing regular updates on progress during the analysis, and 5) developing solution recommendations or implementing solutions.

In most cases, the end user and the OEM have an equal interest to mutually help each other solve a given problem, but times may arise when purchasing agents need to be involved to help move the process along. In the end, all of the proposed solutions may not be feasible, but striving to implement the ones with the best chance of success and those having a positive impact on the equipment reliability is a step in the right direction.

Case Study: PFA Lined Pump Problems

In the production of a corrosive liquid raw material, PFA lined ANSI pumps have been successfully used within Albemarle plant sites for many years, though certain aspects of their performance did not match that of other unlined ANSI pumps.

As data tracking became more prevalent and mechanical seal mean time between failure (MTBF) records were generated, we realized PFA lined pumps had a relatively poor service life. The first generation of PFA lined pumps were severely limited in their reliability in pumping the corrosive liquid due to the high specific gravity of the liquid and the low design margins used for the shaft stiffness.

A redesigned PFA lined pump with a more robust shaft assembly and a more conventional mechanical seal installation was rolled out by the OEM as an upgrade to the existing pumps, with special pricing for replacing their old pumps with the new ones in an attempt to retain their market and phase out the old product line. The new style pump was a better fit for Albemarle applications and did incorporate some design features that Albemarle had previously expressed as needing improvement.

Initial Failure Mode

A test pump with the new design was installed in a service that subjected it to fairly high differential head to help ensure the robustness of the design. After several months in service, the pump was pulled out and inspected. The mechanical seal was in good condition, but some rubbing was noted on the nose of the closed impeller. The rubbing was also seen on the pump bowl in a similar location that indicated contact between the two parts. The rubbing contact inside the pump bowl is shown in Figure 5.

Figure 2While the impeller was being handled after cleaning, it still gave off an odor of the residual chemical from the process. This was unusual since there should have not been any pockets in the impeller where the fluid could get trapped. By accident, when the impeller was tipped on its side, some residual fluid drained from one

of the balance holes drilled through the back side of the impeller. Upon closer inspection, a small recess in the PFA liner inside the balance hole was found. Figure 6 shows a close up picture of the typical damage seen in the balance holes.

Figure 3

 

 

 

 

 

 

 

 

A new impeller was placed inside the pump and the damaged impeller was kept for further analysis. The impeller was thoroughly flushed to remove all residual process fluid. The cleaned part was cut into sections to reveal the source of the defects. The cross-sectional cut showing the balance hole, PFA liner, and stainless steel core is seen in Figures 7 and 8.

Figure 4Lastly, small blisters were noted on the rear pump cover where the mechanical seal gland is attached and seals against the pump casing. The operating conditions were well known for this service, and since this was a new pump installation there were no maintenance issues to investigate other than ensuring the pump was properly installed initially.

Working with the OEM

Now that all the failure modes were identified, a meeting with the local OEM supplier was held at the plant to review the findings. The OEM lead engineer was then contacted and the information was sent for review. The Albemarle purchasing agents were involved at this point due to the nature of the failure and the warranty claim on the parts for OEM defects.

Figure 5After several rounds of discussions between both Albemarle and the OEM, the rubbing contact was resolved by simply adjusting the impeller setting procedure in the field, and by the OEM slightly modifying the dimensions and manufacturing tolerances to ensure proper internal clearance of the pump impeller. A corrective action list was created to keep the team on track with the remaining issues around the recesses in the impeller balance holes.

The issue of the recesses in the PFA in the balance holes was not adequately appreciated by the OEM since there was not a perceived problem with the PFA molding process. In order to ensure quality parts were sent to Albemarle, visual quality checks and spark testing of the lining for pinholes was required for all new impellers.

Over the course of several months, as new pump installations were brought into service, more impellers were caught at the local OEM supplier that did not pass the QC inspection. The matter reached the point where the supplier stock did not have any acceptable impellers, and an unexpected failure requiring an impeller replacement could have shut the plant down while waiting for an acceptable part.

Another meeting was held with the OEM lead engineer to express the need to address the root causes of the defects found in the balance holes. Following many other lengthy discussions and emails, a new manufacturing design was proposed to address the causes of the recesses. After agreeing to the changes, the OEM and their part supplier took several more months to make the first test impellers. Samples were sent to Albemarle for inspection, and one sample was sectioned by the OEM to ensure that no voids or recesses were present.

The new impeller design proved successful after further testing in the plant at Albemarle. The corrective action list was updated with all of the information learned up to that point resulting in a successful impeller design. We requested that the same modifications be made to the other pump impeller sizes so they would not experience the same phenomenon. The OEM did not quickly respond in agreeing to make the changes.

At this time, we decided internally to seek other pump alternatives through other technologies, such as mag-drive pumps, since there were several successful installations of mag-drive pumps in the same corrosive liquid service. Upon deciding to seek other alternatives for pumps, the Albemarle purchasing group notified the OEM there would be no further upgrades of their pump line since they were not supporting the needed improvements in their own product line.

A new product manager at the OEM was taking the reins of the PFA lined pump and apparently did not appreciate the problems or the historical difficulties with their older product line. Finally, after several more meetings with the OEM's management team and the Albemarle corporate purchasing team, the new impeller design was rolled out to the rest of the product line by the OEM.

Short Lived Success with the OEM

As the dust cleared from solving the root causes of the impeller problems with the recesses in the balance holes, it became apparent that the mechanical seals on the pumps were also having issues. In one instance, the pump completely stopped pumping. When it was torn down and analyzed, the entire inner metal core was corroded away by the process fluid.

After examining the possible causes of failure, the O-ring to shaft seal was suspected as the main area of concern. Other pump impellers were inspected and found to have a large gap between the shaft and the PFA lining shoulder at the bottom of the O-ring groove. The gap was so wide on some impellers that the O-rings supplied by the seal vendor would simply slide down between the shaft and the lining into the gap, thereby compromising the seal and allowing corrosive fluid to reach the SS impeller core.

The gap was determined to result from PFA solidification shrinkage, or drawback, following injection molding of the impellers. See Figure 9 for a close up of the gap.

Figure 6 Another meeting was held with purchasing, the OEM engineer, and the OEM product line manager to review the O-ring groove defect. They stated that the primary mechanical seal joint was intended to be between the non-metallic shaft sleeve and the PFA lining on the back of the impeller - even though this was not a proven design. Albemarle did not accept this response due to the lack of flatness and typical mold surface irregularities present, and began rejecting any impeller that did not pass a thorough visual inspection. After several more discussions, the OEM agreed that an improved impeller design was needed to allow the O-ring to function as intended.

Molding modifications were made that eliminated the gap between the PFA lining and the shaft, so that a positive seal could be made between the O-ring and its mating surfaces. See Figure 10 for the improved design, which yields a zero gap between the lining and the shaft.

Figure 7The new O-ring seal design was implemented and has been successful in eliminating that failure mode for this pump design in the plant. All was quiet again with parts issues from the OEM until another strange failure mode appeared in the form of large divot marks (i.e. rubbing wear damage) on the back of the PFA lining on one pump impeller. Figures 11 and 12 show the details of the divots found on the back of the impeller.

Figure 8More meetings have been held and discussions on corrective actions are ongoing with the OEM. To date, no significant concrete changes have been proposed by the OEM to address any of the suspected root causes for this newly observed defect. Evidence thus far suggests this defect is the result of thermal deformation and "lifting" of the PFA lining, which spans the five "fingers" of the metal core. In this regard, the design flaw may be a combination of inadequate stiffness or inadequate restraint of the PFA lining on the back of the impeller, combined with overly tight impeller-to-housing clearances.

Figure 9In this context, the result is a product that does not have a rugged or tolerant operating envelope. Thus it may work acceptably in many services, but fail repeatedly in others. The pertinent point here is that this type of failure may occur solely as the result of design "frailness" and not from meaningful process or operational nuances.

Summary

With this OEM, many challenges have been overcome, deficiencies eliminated, and improvements made to the product for both Albemarle's and the OEM's benefit. Mechanical seal life has steadily improved over the last two years for this new PFA lined pump design as compared to the older pump design.

However, it is crucial to continue to seek other alternatives for pumps in this application if the pump and mechanical seal reliability and parts quality of this OEM's products do not approach that of other ANSI pumps in the rest of the plant.

Special thanks go to all those who have been involved with investigating, researching, and driving solutions for the above problems, specifically Mr. Wayne McQuiston - PM/Pd Crew Leader, and the PM/Pd Maintenance Crew; Mr. Kenny Kendrick - Transportation Supervisor, and the transportation equipment maintenance crew. Without their persistent daily efforts these types of problems would possibly remain hidden and unresolved.