A reliability-centred framework for sub-asset management in CMMS

Over the course of managing maintenance programs in oil and gas operations – particularly within stimulation and coiled tubing fleets – I began to notice a recurring pattern. Engines, transmissions, power ends, and fluid ends degraded at different rates and failed for different reasons. Their operating exposures varied significantly depending on duty cycle, load profile, and environmental conditions. Despite this mechanical reality, most CMMS configurations treated these components as attributes of a single parent asset. All failure events, work history, and meter accumulation were consolidated at the unit level.

From a reporting standpoint, the system appeared organised and complete: dashboards populated, work orders closed, and preventive schedules generated on time. Yet from a reliability standpoint, the structure concealed the very information required to improve performance, because the degradation behaviour of individual components was never isolated or tracked independently.

When work history is aggregated at the equipment level, component-specific failure behaviour becomes diluted and difficult to analyse. Preventive maintenance then defaults to calendar-based intervals, not because it is the best strategy, but because the system does not capture exposure where degradation actually occurs. Root cause analysis tends to focus on restoring the asset to service rather than understanding how individual components behave over time, and as a result, the same failure modes often recur without meaningful change.

Early in my career, this configuration seemed practical. It simplified hierarchy setup, reduced administrative overhead, and aligned with legacy CMMS capabilities that did not fully support serialised sub-asset management. For smaller fleets operating at moderate utilisation, parent-level tracking was adequate, and expected life could be estimated using broad historical averages.

As utilisation increased, the structural limitations became increasingly difficult to ignore. Maintenance costs began rising in ways that could not be attributed solely to asset age or workload, and major components were removed, rebuilt, and reinstalled across multiple units without preserving their accumulated exposure history. Engineers compensated by maintaining parallel spreadsheets to track engines and fluid ends manually, because the CMMS could not follow those components through swaps, and overhaul timing gradually drifted away from actual degradation patterns. None of this reflected poor data discipline; it reflected a hierarchy that failed to mirror the physical behaviour of the equipment itself.

Over time, it became clear that failures were being evaluated at the wrong level of abstraction. Complex equipment does not fail as a monolithic unit; it fails through the behaviour of its constituent components, each of which experiences different stressors and degradation mechanisms. Treating the parent asset as the primary reliability entity masks this variability and, in doing so, obscures the real drivers of maintenance cost and repeat failure.

Reframing the problem: a field example

Consider a stimulation pump operating in hydraulic fracturing service. These units run at high utilisation and experience frequent component replacement. Engines accumulate runtime under varying load conditions, fluid ends experience cyclic pressure fatigue, and transmissions degrade under fluctuating torque loads, all within the same operating unit. Each component responds differently to stress, yet the original CMMS configuration treated the pump as a single reliability entity.

Under the original CMMS structure, the pump was configured as a single parent asset, with all operating meters and maintenance logic tied to that top-level record. Preventive maintenance triggers were therefore driven by total unit runtime rather than by the actual exposure of individual components. When an engine was removed for overhaul and installed in another unit, its accumulated exposure history did not travel with it, meaning the receiving pump effectively inherited the engine’s future failure risk without any visibility into its prior operating conditions. Over time, this disconnect distorted reliability analysis across the fleet, because component behaviour was being interpreted through incomplete or misaligned data.

Once the hierarchy was redesigned, each critical component – engine, transmission, power end, and fluid end – was established as a serialised sub-asset rather than an attribute of the parent unit. In practical terms, this meant that every component now carried its own installation date, configuration attributes, and exposure tracking, allowing lifecycle data to accumulate independently of the equipment in which it was installed. Engine runtime, for example, was cascaded from the parent asset through controlled meter relationships so that exposure accumulated automatically without creating manual duplication or parallel tracking. With that structure in place, preventive and condition-based tasks could be assigned directly to the component responsible for the failure mode, whether that involved OEM service intervals, oil sampling protocols, or inspection requirements tied to specific degradation mechanisms.

When engines were swapped between pumps, their full histories – runtime, overhaul records, failure events, mean time between failures (MTBF), and remaining useful life estimates – moved with them. For the first time, the fleet could be viewed as a collection of managed components rather than interchangeable units. Forecasting rebuild demand became more accurate because life consumption was calculated where wear actually occurred.

Designing a reliability-centred sub-asset framework

The framework that emerged from this work was not theoretical; it evolved from repeated field failures and the need to understand them more precisely. At its core, it reflects a simple but consequential principle: reliability decisions must be made at the level where degradation mechanisms operate, not at the level most convenient for reporting.

Sub-assets should not be defined for accounting convenience or reporting simplicity, but according to the way components actually fail in the field. Each sub-asset must have clearly identified degradation drivers, independent exposure accumulation, and its own maintenance and failure history so that engineering decisions can be tied directly to observable behaviour. Serial continuity is equally critical, because lifecycle data must remain intact through removal, rebuild, and reinstallation if reliability modelling is to remain credible.

Hierarchy design must mirror physical cause-and-effect relationships, otherwise analysis becomes disconnected from reality. Corrective, preventive, and condition-based work orders should be raised against the component that failed rather than simply against the parent unit, and failure modes and causes should be assigned at that same level. When structure aligns with mechanics, reliability analysis reflects true component behaviour instead of reporting abstraction.

Exposure modelling deserves particular attention, especially in high-utilisation fleets. In many CMMS environments, runtime is recorded only at the equipment level, which forces planners to infer component wear indirectly. By implementing meter cascading and attribute mapping, relevant exposure variables – such as runtime hours, load severity, and pressure cycles – can propagate automatically to sub-assets. This approach preserves data fidelity without increasing administrative burden and ensures that wear is tracked where it physically occurs.

Beyond exposure tracking, lifecycle continuity becomes especially important in high-utilisation fleets where components are frequently removed, rebuilt, and redeployed. Serialised sub-assets must retain complete historical records even after overhaul, and when reinstalled, they should resume accumulation from clearly documented baselines. Rebuilt components may begin new accumulation cycles, but only with defined reset logic that preserves prior exposure context. Over time, this continuity enables statistically meaningful reliability modelling, replacing anecdotal interpretation with defensible engineering insight.

From calendar-based maintenance to exposure-based planning

Traditional preventive maintenance strategies rely heavily on fixed intervals, which assume uniform degradation across operating conditions. In practice, however, engines degrade according to runtime and thermal load, fluid ends respond to cyclic pressure and abrasive wear, and control systems deteriorate under vibration and environmental stress.

By establishing baseline expected life using historical performance data and engineering design limits, and then adjusting that baseline according to actual exposure severity, maintenance planning shifts from assumption to quantification. Abnormal events such as overload conditions, contamination, or sustained high-temperature operation can be factored into the model as life penalties, while improved maintenance practices and controlled operating conditions may extend usable life beyond nominal expectations. In this context, remaining useful life is no longer a static estimate derived from calendar intervals, but a continuously recalculated measure that reflects real operating behaviour and accumulated stress.

This approach does not remove uncertainty from maintenance planning, but it does reduce it to a quantifiable and defensible range, allowing leadership to make decisions based on exposure-driven evidence rather than broad historical averages.

Operational and financial outcomes

After implementing serialised sub-asset modelling across the stimulation fleet, failure forecasting improved materially. Rather than approximating overhaul timing based on broad historical averages, engines and fluid ends were scheduled for service according to accumulated exposure and documented degradation history. Procurement planning aligned more closely with predicted failure windows, which reduced emergency orders, stabilised inventory requirements, and minimised idle capital tied up in reactive replacements.

In 2025, this structured approach contributed to approximately US$800 000 in savings for stimulation pump operations through optimised overhaul scheduling, improved spare inventory alignment, and a measurable reduction in repeat failures. Equally important, maintenance cost per operating hour declined, warranty tracking became more precise at the component level, and reliability improvement initiatives shifted from reactive troubleshooting to targeted engineering interventions.

For example, engine reliability programs could focus on specific degradation mechanisms – fuel injector wear, turbocharger thermal stress, filtration performance – because failure data was captured at the correct level. Improvement efforts became data-driven rather than reactive.

Reliability as a system of components

Viewing reliability as the aggregate behaviour of individual components allows engineering teams to analyse performance with far greater precision. Within that structure, failure events can be correlated directly with exposure severity, premature degradation patterns can be isolated, and maintenance-induced variability can be identified and corrected with greater confidence. As component-level behaviour becomes clearer, overall asset reliability can be derived using established statistical models, creating an analytical foundation for capital planning and operational risk management.

Organisations that adopt this approach do not eliminate maintenance variability, but they gain meaningful visibility into the factors that drive it. When exposure history replaces broad historical averages as the basis for forecasting, planning confidence increases because projections are anchored in documented component behaviour rather than generalised assumptions. Capital allocation decisions become more defensible as well, since individual component lifecycles are traceable, measurable, and supported by accumulated operating data.

Sustainable reliability improvement begins with how assets are structured inside the CMMS. When system hierarchy reflects physical reality, data becomes analytically useful rather than merely reportable. This process improvement generates more meaningful outcomes: maintenance decisions become more precise, forecasts gain credibility, and operational performance is optimised because planning is grounded in component-level behaviour rather than aggregate assumptions.

Maintenance organisations often invest heavily in analytics, dashboards, and predictive tools, expecting insights to emerge automatically from accumulated data. In practice, those tools deliver value only when the underlying asset structure reflects mechanical reality and component-level behaviour is preserved. When CMMS hierarchy is designed with that intent, data becomes actionable rather than archival, planning becomes disciplined rather than reactive, and performance improvements become durable rather than temporary. For organisations seeking measurable gains in uptime, cost control, and asset longevity, the most consequential reliability decision may not be which technology to deploy, but how the system is structured from the beginning.

About the author

Rajaram Madhavan is an engineering leader with nearly 20 years of experience at SLB, where he applies his expertise in mechanical systems, asset reliability, and data-driven performance optimisation within oil and gas operations. Over the course of his career, he has held leadership roles in mechanical design engineering, product development, and reliability support. He currently serves as Manager, Master Data and Business Systems, based in Sugar Land, Texas.

Raj holds a Master of Science in Mechanical Engineering from the Georgia Institute of Technology and is completing a Master of Science in Data Science at Indiana University Bloomington.

The views expressed in this article are his own and do not necessarily reflect those of SLB.

Read the article online at: https://www.oilfieldtechnology.com/special-reports/07042026/a-reliability-centred-framework-for-sub-asset-management-in-cmms/