Presented by:

E&P Plus

Editor's note: This article originally appeared in the April issue of E&P Plus.
Subscribe to the digital publication here.

For an oil and gas asset management company, maintaining and inspecting plant and associated equipment efficiently can often feel like a never-ending process. Whether it is contending with budget, time or bed space constraints, integrity teams are far too familiar with assessing, prioritizing and justifying their decision-making processes. In highly regulated locations such as the U.K. Continental Shelf, not taking the time to properly assess and understand an asset condition can be an exceptionally expensive gamble, which is why most operators pursue a risk-based inspection (RBI) methodology. But is this approach still fit for purpose in 2021?

Good quality RBI has its place in the asset management tool kit as it enables direct comparison of all the equipment under the metric of risk, irrespective of their damage mechanisms. For example, asset managers can use RBIs to compare the risk of corrosion on a piece of pipework against the risk of fatigue cracking to the deck integration on the skid for a pressure vessel. By incorporating feedback, such as material or equipment condition into the asset’s profile, the recommendations will change over time. But it is this feedback loop that can sometimes be the undoing of an otherwise sound asset integrity methodology.

The creep of human bias

Anecdotally, the industry has seen how human bias creeps into RBI strategies over time. Engineers lose confidence in the inspection readings and err on the side of caution. After a few inspections, this shifts the asset’s risk profile further away from where it should be statistically. The profile becomes overly conservative, assessing probability based on engineering experience and past scenarios to the point of being toxic to the ethos of the very approach it was designed to deliver.

In other instances, the RBI strategy is primarily based on whether an incident has occurred within an organization or is a common issue within the industry, and it will only ever be based on historical or assumed data rather than what is happening in real time. Equally, much of the decision-making is made in isolation of what is happening in the plant’s wider operations, meaning valuable operational insights, such as process data or system changes, are not incorporated. All in all, it is fair to say there is room for improvement in what is currently one of the most common integrity management processes used in heavy industries.

Holy Grail of asset integrity

Event-driven inspection based on predictive integrity management is the new Holy Grail for many asset managers. Rather than using traditional lagging indicators such as wall loss or visible corrosion, the responsibility is on using leading, real-time (or very near-real-time) data, such as process conditions including temperature and pressure, to drive inspection.

As an example, under one set of conditions, pipe X will degrade at a certain rate per annum. Those conditions are monitored using environmental and process data and log any events where these data deviate from the model’s “business as usual” circumstances. At a given point in time, an alteration is made concerning the production fluids within the system, causing an increase in pipe temperature by 20 C. Because higher temperatures accelerate corrosion, this creates an event that must be accounted for and reviewed.

Modeling environmental and process data in this way improves accuracy and efficiency of inspection and intervention. By creating a data model based on the conditions and fluctuations that affect degradation rates and monitoring these conditions accordingly, operators can confidently improve investigative time to those components where conditions have changed.

Event-driven inspection is a method that is particularly relevant for scenarios where degradation mechanics and parameters are well understood and particularly beneficial for complex operations where the cost of traditional inspection methods add up quickly. If an operator can confidently monitor condition and only command inspections when absolutely necessary, substantial savings can often be made. However—particularly in older assets—the reality is sometimes a little less black and white.

There will always be risk-based integrity activity, as even the most accurate model will acknowledge bounds of error and predicted condition will begin to diverge from known condition over time at a rate based upon quality and availability of information. With event-driven inspection, asset managers will be concerned with curating/managing and validating their asset datasets.

One of the current outputs of IMRANDD’s analytics software is that it identifies gaps in datasets in addition to predicting failures and identifying potential savings. These gaps can be viewed as blind spots. Identifying them is a good step in itself, which is moving from “unknown unknowns” to “known unknowns.”

Gaining access to real-time data

Most operators have been collecting real-time data on asset operations for decades through sources such as sensors, production and chemical data, to name a few, but not actually using it in real time. However, recent advancements in data collection and analysis make this new Holy Grail of event-driven inspection possible.

First, the ability to use additional real-time datasets captured by different departments make predictive modeling more targeted and build up a better, more accurate and holistic view of an asset’s condition.

Secondly, data no longer need to be manually curated before an onshore engineer attempts to try to interpret the detail. Instead, data are being screened, analyzed and validated continuously, providing the timeliest view of an asset’s integrity and substantially reducing time to implement asset management decisions.

Finally, new software exists that can build complex models and analyze data separately helping to eradicate human bias entirely. Moreover, this approach does not require any hardware, simply an ability to think differently about the types of data from an asset that could be used to provide insights for the integrity management team.

Confidence in greater volumes

This last development is particularly important for event-driven inspection where the ability to handle large volumes of data seamlessly and sort errors from genuine data variations is a prerequisite for having confidence in recommendations. While many variations can be legitimately disregarded—be they from human error, a poorly calibrated kit or prevailing conditions—a handful will be legitimate indicators of failure. Operators need to be certain of which data can be excluded, which inconsistencies require further investigation and where gaps lie.

IMRANDD’s forthcoming AIDA MAX will be one of the digital asset management solutions that can do just that, using machine learning to assist the build of its asset degradation models to power event-driven inspection. Built for engineers, by engineers, the new analytics software will bring critical, real-time or near-real-time insights to complement existing human checks and balances.

With this technology, data variations are separated out from the normal ranges to be analyzed with more scrutiny and eventually reincluded in models. Furthermore, given the volumes of data being handled, AIDA MAX will be able to identify and treat many data variations automatically without an engineer’s intervention.

As example, where the data show pipe X’s process conditions have changed, the model is then updated with this new information, changing the degradation model that now indicates a likely failure prior to the next planned intervention. The plan is updated based on this new leading indicator, which in turn informs the integrity team to inspect the affected area sooner.

While using historic data to make quantitative assessments has been a staple method for asset integrity management for some time, the benefits of utilizing real-time data are undeniable, and the insights harnessed continue to improve rapidly in line with the industry’s growing digital capabilities. Operators stand to gain a more holistic view of their plant’s operations and condition as well as greater confidence in their asset integrity management. Ultimately, this will lead us to adopt event-driven integrity and inspection where continuously updated degradation models allow operators to precisely predict integrity and make stronger commercial decisions.

Offshore