Developing or procuring appropriate software is crucial. The important questions to answer are:
• Does your site possess the means to capture accurate causal effect data from multiple sources in a meeting environment while maintaining efficient use of the employees' time?
• Once captured, is the information easily transmitted across work groups within the facility or the corporation, if necessary?
• Is there a working action-item system that allows cradle-to-grave tracking of such items?
CONDUCTING AN RCA
Following an incident, a sequence of orchestrated events should occur prior to the RCA meeting. Unit and compliance supervision typically lead an initial field investigation. The site's PSM guiding documents then would trigger the initiation of an RCA. At Lake Charles, this involves notifying the specialist to formally begin the RCA process. A source event is created in our electronic tracking system, documentation is gathered and a meeting is scheduled.
A well-organized and consistently facilitated meeting producing effective results can lessen the potential for attendee frustration (Figure 2). Aside from weak meeting facilitation, the primary pitfalls of typical RCA meetings arise from two failings — lack of representatives of key constituencies in the group and lack of necessary information.
Consider an incident owned by a particular production unit. While it may seem sensible to limit the RCA meeting to its production staff, this ignores the impact the unit has on internal suppliers, such as maintenance and safety, and customers, such as downstream operations, as well as external suppliers and customers. The absence of a necessary, but overlooked, team associate may undermine the results of the meeting.
Although the unforeseen may occasionally occur, it is disconcerting to have a meeting run into a roadblock or slow down due to lack of necessary information, including procedures, visuals, failed parts or incident scene photographs. Clearly, the better job an incident owner does in preparing information for the RCA meeting, the more effective and consistent the meeting. Of course, an unforeseen need may arise.
At Lake Charles, we have markedly improved preparatory work through two primary means: a standardized RCA Scheduling Request form and a pre-meeting.
The form aids incident owners in assembling the proper team. It also serves as a prompt to submit certain types of records critical to the RCA. Each team member's name and documentation entry is provided to the specialist for proper electronic entry into the incident tracking system. Assignments then are delegated to appropriate team members.
The pre-meeting is initiated by the RCA specialist once a full meeting has been scheduled. It is a very brief gathering of one or two key incident owners that focuses on three primary areas: 1) verifying that the proper team members are indeed scheduled; 2) ensuring records items (photos, procedures, etc.) are adequate for what the RCA meeting hopes to accomplish; and 3) developing an initial failure effect (also known as primary effect, significant event, etc.), i.e., a single well-defined focus for the team's efforts. In addition to the failure effect, the pre-meeting also adds primary failure modes to provide a framework that paves the way for smoother management of the actual analysis meeting.
Another trap during an RCA meeting is indulging in the blame game. This threatens the very fabric of RCA consistency. It often leads participants to kick into survival mode out of the fear of negative consequences. To counter this, we stress when starting an RCA meeting that determining the cause, not someone to blame, is the intent — and we repeat that message often during the session. This has proven effective for us and undoubtedly has improved the odds of obtaining honest and accurate answers during construction of any causal diagram. It doesn't mean we won't discipline a person found in clear compliance violation. However that aspect is dealt with independently prior to the RCA meeting. After all, consistently and fairly applied justice can strongly deter unwanted human actions.)
A well-facilitated RCA can be extremely powerful in focusing on both chronic and sporadic incidents. Skillful lines of facilitator questioning, combined with use of suitable software, can guide participants to delve deeper into causal systems at play. In most cases, it's essential to look well beyond the obvious details of the incident. A team's effort should focus on those latent causes found deeper in the internal and cultural systems in which we work and operate.
We further enhance quality and consistency via an approval process managed by the RCA specialist. All applicable leadership affiliated with the incident review the outcome, provide input and give assurance that the primary causes have been identified. Prior to having the specialist manage the approval process, some RCA investigations and reports remained in an open status for months. Operating units could fall into the trap of becoming satisfied as long as an RCA was conducted and action items entered. Once RCA management began, time-to-publish rates decreased dramatically from an average of more than four months per incident to slightly over one month.
Over the past three years, we've given the publish rate of analyses much closer scrutiny, with a shift in focus from "get it published" to "get it correct and then get it published." Hence, publish rates now hover at around the 40-day mark.
One of the key benefits of electronically creating an RCA cause-and-effect diagram (tree) is the ability to link the team's Logic Tree worksheet to an executive report that includes not only a summation of the team's conclusions but also a roster of necessary corrective actions. Our electronic tie-in between reports and accompanying action items perpetually links RCAs, summary reports and corrective actions.
Our software choice allows for generating action items as well as for the subsequent approval process for their completion and verification. Thus, items don't become lost in an electronic wasteland but are tracked to completion, including the verification of whether or not the action taken indeed was effective in preventing the root cause from resurfacing. The verification-of-effectiveness step also offers a safety net for owners taking alternative actions should the proposed corrective action not provide the desired results.
We take several steps to facilitate the sharing of RCA information. During the formal approval process, managers can nominate an RCA report for in-plant sharing. Then, once the RCA is complete and approved for closure, the report is saved and electronically stored into the published (or closed) file of the PROACT system, allowing read-only access. Additionally, compliance team leaders meet to nominate incidents for both plant-wide and corporate sharing.
MAKE THE MOST OF AN RCA
We've found that several elements are essential. You must know when an RCA is necessary, which typically is driven by a documented procedure. Then, you must assemble the correct cross-section of attendees and give them appropriate support documentation. Unbiased and effective facilitation in the meeting remains the tipping point between success and failure. Oversight of RCA quality and quantity is crucial. A structured and strong approval process, coupled with sharing of findings, closes a loop for tying together other incident causes that previously may have been unconnected. You also must verify that action items are completed and effective.
Consistency of the RCA doesn't begin and end with the analysis meeting. An entire process shaped around the meeting event represents the ideal in attaining improved and consistent RCAs.
MICHAEL I. GUIDRY is VPP tech specialist and former RCA specialist for Axiall Corp., Lake Charles, La. E-mail him at email@example.com
1. Gano, Dean L., "Apollo Root Cause Analysis: A New Way of Thinking," Apollonian Publications, Richland, Wash. (1999).
2. Latino, Robert J. and Latino, Kenneth C., "Root Cause Analysis: Improving Performance for Bottom Line Results," 2nd ed., CRC Press, Boca Raton, Fla. (1999).