“The hardest thing is writing a recommendation for someone we know.” — Kin Hubbard
We’ve all heard, over and over, that data without analysis is useless. Well, analysis without recommendations is just as useless. Unfortunately, many young engineers are incredible analysts but then mistakenly believe that their analysis will stand on its own, that the recommendations will speak for themselves.
I recently had an epiphany: young engineers don’t make recommendations because as engineering students, they haven’t been taught how. Most will eventually figure it out, but only after years of doing it badly.
In process safety, particularly in compliance with OSHA’s Process Safety Management (PSM) Standard (29 CFR 1910.119), there are several requirements for recommendations: explicitly in process hazard analyses, in incident investigations, and in compliance audits, and implicitly throughout the regulation.
It’s not enough to make a recommendation, however. To be useful, they must be good recommendations. In 2016, we began writing weekly blogs about safety. To date, we have posted almost 400 blogs. Almost 20% of them were about recommendations. Some were about the mere existence of recommendations. Over half were about specific recommendations. The remaining third, though, were about the purpose of recommendations and how to meet that purpose.
What follows is the evolution of my approach to recommendations.
As A Young Engineer
Some of my earliest recommendations were trivial.
“Don’t do that”
is about as bad as it ever got. It assumed that the reader of the recommendation knew what I was talking about, that they had immediate access to all the same documentation that was then at my fingertips and shared all the thoughts that were then front of mind for me.
As for incident investigations, this is the worst recommendation I ever produced:
“Repair this”
If it is not obvious, the problem with both of these recommendations is that the person responsible for executing the recommendations had no idea what they were. Once they tracked down what the recommendations were asking for, it became clear that neither recommendation reduced risk. “Repair this” restores the status quo but does nothing to make a recurrence less likely. And “Don’t do that” gives no clue about to whom it applies.
As A Young Engineer With More Experience
In response to some early criticism, I got better, especially about being more specific. An example of an improved recommendation was
“Install a PSV on Solvent Filter, F-101”
The person assigned responsibility for this didn’t have to go looking to discover what they were supposed to do. On the other hand, it gave them no flexibility to develop a better solution. It lost sight of the concept that while the PHA team may be well qualified to identify hazards with risk that was too high, they were not necessarily the best to identify the solution.
The recommendation didn’t give more detail than that. Was the relief valve for thermal expansion, the blocked in and external fire case, or something else? Did it need to be on the housing itself, or could it be on inlet or outlet piping? My manager at the time convinced me that the recommendation didn’t need to address all the details, that the PHA was not the time or place for detailed design. Instead, I could rely on the person being assigned to that task to be qualified to develop an appropriate design.
As An Engineer With Even More Experience
Later, that same engineering manager shared with me their experience of being at a facility that OSHA cited for not implementing a recommendation as it was specified by the PHA team. Instead of a PSV, they decided to install a rupture disk. Then, during the design, the project engineer discovered that the filter housing had an inside diameter of less than six inches, meaning that codes treated it as a piping accessory, not a pressure vessel, and that it did not require pressure relief.
To address that lack of flexibility, I began structuring recommendations differently:
“Consider installing a PSV on Solvent Filter, F-101”
Some objected, however, to the use of the word “consider.” They saw it as a weasel word, a word that would allow managers to review the recommendation and then disregard it. “Yep, I considered it and we’re not doing it.” To counter this, I made sure that PHA reports included this statement:
Items to consider are typically those items that would act as additional safeguards against the hazard being reviewed. The recommendation is to evaluate the unmitigated risk of the hazard and then to compare it to the mitigated risk achieved after taking credit for the existing safeguards. Should the mitigated risk be greater than the risk that [company name] is able to tolerate, the evaluation should then go on to consider whether the suggested safeguard would be effective in reducing the specific risk of the hazard as well as the overall risk of the process, whether alternatives not recognized by the HazOp team would be more effective, and what the plan should be for implementation. A recommendation to consider a safeguard should not be treated as an absolute recommendation to implement it or preclude the development of alternate solutions to address the hazard.
OSHA requires that every recommendation developed during the PHA of a PSM-covered process be resolved and documented, even if the resolution is something other than implementing the recommendation as written in the PHA.
As An Experienced PHA Facilitator
With the passing of time, I wanted to phrase recommendations less passively. I still liked using the word “consider” for the flexibility it implicitly granted, but I wanted it clear that something must be done. So, instead of focusing recommendations on a proposed solution, a solution that may or may not be the best approach, I began writing recommendations that focused on the problem:
“Consider how best to provide Solvent Filter, F-101, with adequate pressure relief that complies with the ASME BPVC, Sec VIII, and implement”
Just because a PHA team isn’t necessarily the best qualified to develop solutions, that doesn’t mean they are without a clue. During a PHA, they focus on each node and may, in fact, generate great ideas. It’s important to capture those ideas in the recommendation without constraining the final resolution. So, I sometimes use a variation similar to this:
“Consider how best to provide Solvent Filter, F-101, with adequate pressure relief that complies with the ASME BPVC, Sec VIII, e.g., installing a PSV sized for the fire case on the outlet piping upstream of the first block valve, and implement”
As An Experienced PHA Facilitator in a Post-STAA World
I’m still evolving. My most current iteration in formatting recommendations is a response to the EPA’s recent requirements for Safer Technology and Alternatives Analysis (STAA). We are starting to add a parenthetical comment at the beginning of each recommendation at the time it is developed, identifying whether the recommendation qualifies as “substitution,” “minimization,” “moderation,” “simplification,” “passive,” “active,” or “procedural.” Then, once we generate the list of recommendations, we can easily sort by type of safer technology and alternative in order to satisfy one of the EPA’s new requirements. In the case of the filter housing pressure relief, it would look like this:
“[Active] Consider how best to provide Solvent Filter, F-101, with adequate pressure relief that complies with the ASME BPVC, Sec VIII, e.g., installing a PSV sized for the fire case on the outlet piping upstream of the first block valve, and implement”
An Approach
A few years ago a couple of consultants from Calgary presented a paper at the 16th Global Congress on Process Safety where they adapted Doran’s original S.M.A.R.T. approach to goal-setting to process safety. While there were details I disagreed with, I thought they had suggested (recommended?) a great overall approach to PHA recommendations:
Stand-alone
Measurable
Assignable
Risk-reducing
Trackable
If you have gotten this far, I recommend (there, I said it) that you read both their paper and my related blog.
Finally, coach the young engineers who are fresh out of college that you work with. They’ve learned how to analyze but have probably not been taught how to transform their analyses into recommendations. It will come with experience, but it doesn’t have to be theirs. Let it be yours.
Leave A Comment