We’ve been discussing Scalable Accuracy and its use related to Lifecycle Management technologies available to owner/operators. The last few topics have included Fitness for Service using Scalable Accuracy, the approach for Equipment Lifecycle Management and, to lay the foundation for proper thinking, making the case for Scalable Accuracy. In a continuation on the topic, this week’s blog will cover Risk-Based Inspection (RBI).
When using the scalable accuracy approach, we typically start with RBI. It is a “higher level” approach, allowing us to perform initial risk screening on literally tens of thousands of pieces of equipment, risk ranking them, in relatively short order.
Our examples will be based on API RP 581 of which the latest edition is very scalable. While it is semi-quantitative to quantitative, the quantitativeness of it is based on the following:
- Level of damage mechanism review:
- The PFD level review is, generally speaking, the most coarse and therefore less quantitative, in other words more semi-quantitative;
- The P&ID level review is more quantitative;
- Input data (Actual, accurate data produces results with less scatter, hence more quantitative);
- Where actual data is not known or is very expensive to obtain it may be acceptable to use assumptions. Assumptions must always be made by knowledgeable people, as is appropriate for the assumption. For example, if corrosion rates as estimated by measured field thickness readings are unreliable the analyst may want to use a corrosion engineer’s best estimate. The corrosion engineer should be very familiar with that type of process unit and equipment, consult with operations to learn about operations as they relate to damage and expected damage rates, get the input of the area inspector, take into account similar types of units in industry, consider past failure histories for that unit and similar industry units, apply chemical, materials and physical laws/rules, etc. to arrive at a reasonably conservative corrosion rate for the analysis and future planning. The more one uses reasonably conservative assumptions the less accurate the analysis will be, erring to the conservative. This will often be more advantageous than conventional approaches.
The conservative inputs may eventually drive the owner operator to get more accurate data to drive the probability of failure down, if risk justifies it. For example;
A pressure vessel was commissioned into service in 1997. Original thickness was 0.500”, SA 516-70 material. An industry representative corrosion rate of 10 mils per year was used for the original RBI analysis. The team felt the real corrosion rate was closer to 5 mpy but wanted to be conservative. The RBI analysis shows the risk target or threshold being reached on April 5, 2014. The thinning mechanism is localized. RBI is calling for an A level (The inspection methods will correctly identify the true damage state in nearly every case, or 80–100% confidence) effectiveness inspection for localized thinning. So on or before April 5 a “B” scan of 75% of potentially corroded area, as specified by a corrosion specialist, will be performed. This will cost money to get more data, to become more accurate, and risk says this is justified. Returning data confirms the actual corrosion rate is 5.5 mpy. The vessel operated safely for 17 years. The actual metal loss for those 17 years was 17 X 5.5 = 93 mils. So now the thinnest area on the vessel is 0.407”. The analyst may exercise the option to change the corrosion rate for analysis to something less than 10 mpy, as well, for the next calculation. This is another example of scalable accuracy and done at a higher level than FFS.
To take this example a step further let’s say that the current t-min is 0.400”. If risk justifies, an FFS may be performed for the future. The user may progress from a level 1 to a level 2 FFS as the calculated t-min from the 581 analysis may be near or the same as the design t-min. We will pretend the 579 level 2 calculated t-min is 0.300”. It will be important to notate each item in the database with t-min’s and resulting PoF calculations that are based on a FFS analysis versus a more conservative design or API RP 581 based t-min. When the required ligament or t-min is dropped to 0.300” the PoF will decrease, the calculated future wall loss rate will drop based on the confirmed 5.5 mpy corrosion rate, the future PoF will decrease compared to what it would have been with an 0.400” t-min. As the vessel wall continues to grow closer to the t-min risk will rise again. Then options such as de- rating, etc. might be exercised to keep the vessel in service. This example also shows the relationship between RBI and FFS.
It is important to note that for both RBI and FFS (heavily dependent upon the FCA or future corrosion allowance) process changes, other changes such as those handled by your MOC program, anything that can impact the damage rates and types of damage that the equipment can experience, must be monitored so we know if and when they occur. This will allow us to consider the information, reanalyze if necessary and make any necessary adjustments to our plans.
As mentioned in the FFS blog post, it is also important to note that these three levels are calibrated against one another and do not conflict with one another as long as they are performed properly as per the procedures in the referenced standard. The same is important for RBI analyses. If one is using tools with discretely different methods for quantitative and semi-quantitative analysis, they must be calibrated to each other so they produce results do not conflict.
For example, I recall a population of 200 pieces of equipment that were fully analyzed at both level 1, qualitative, and level 2, semi-quantitative RBI. The original intent was to perform an initial RBI of the 200 items qualitatively and move high and medium-high risk items to the semi-quantitative analysis. Some of the equipment analyzed at the semi-quantitative level were rated at a higher risk than they were at the qualitative analysis level and should have been graduated to semi-quantitative, but never would have if the original strategy was implemented. The beauty of the 581 approach is that all equipment items are analyzed on the same platform and we are comparing ‘apples’ to ‘apples’, i.e. they are calibrated. The qualitativeness is introduced, at the desired level, via the approach and the amount of assumptions used. As more accuracy data is needed, it is added, as dictated by risk, and all equipment is being risk ranked on the same level playing field.
Next week will be the wrap up on Scalable Accuracy. Be sure to sign up for our email newsletter so that you get notified. As always, you can visit our LinkedIn Group page to post comments on this and other topics.