Predictive Maintenance for Aging Oilfields – Introduction
This is the introduction article to a series of articles about investigating the effects of Predictive Maintenance on oil production for aging oilfields (“Brown Fields”). I will use MS Excel and the open source statistical language R for building Monte Carlo and Machine Learning Models to answer the question “does it make economic sense to invest in predictive maintenance for a given oilfield?”
Oilfield Production Profile
Aging oilfields are old fields that are still producing, but at a declining rate and with ever increasing maintenance costs. At some point of the field life, operating costs will exceed generated revenue and the field will be decommissioned. The decline phase is usually the longest phase in the field life cycle, as shown below.
Single Well Production Profile
An example of the qualitative production rate of a single well is shown in the figure below. At the start of production, the well is at its peak production rate, followed by an immediate steep decline.
Often the production rate is displayed on a logarithmic scale, showing the decline as a straight line.
Effect of Maintenance Events on Oil Production
Maintenance costs make up a large part of operating costs; the older the equipment, the larger the wear-out failure rate and the more often repairs are needed. This means higher inventory costs to keep spare parts ready, a sizable maintenance department, additional logistic costs and more non-productive time; most wells have to shut down production during repairs and must be started up again later (not a trivial flip of a switch for most wells!).
During the maintenance period, often the well is shut down, resulting in non-productive time. Once repairs are completed, the production is started up again; ideally, the production rate will reach the same rate as predicted by the decline curve. The total production loss is the rate per time multiplied by the shut-down time, plus the period of lower production during the startup phase.
Early failures may occur almost immediately, e.g. because of manufacturing defaults or repair/installation mistakes. Random failures can happen at any time, and wear out failures are mostly mechanical failures that happen because of aging equipment.
Current Maintenance Strategies
The most common maintenance strategies used today are:
- Breakdown Maintenance
- Preventive Maintenance
Breakdown maintenance means that repairs are done once a failure occurs. Depending on how critical the replacement parts are, they might be kept available in the spare parts inventory or not. Usually such events come unexpected and maintenance resources must be added to the existing maintenance plan.
Preventive maintenance means scheduled inspections and services to prevent the equipment from failing (like regular oil changes in a car). Because these are scheduled events, they are already included in the existing maintenance plan.
Predictive maintenance is a technique that uses accumulated equipment data (sensors, inspection, etc.) and a failure predicting model to forecast equipment failures so that repairs can be scheduled before the failure actually happens.
Why is this approach attractive? Because it is more efficient and cost effective than other maintenance strategies. Some of the advantages are:
- Less non-productive time
- Lower inventory cost
Less Non-Productive Time
Unexpected breakdowns, even if detected immediately, will cause additional non-productive time for an oil well because the maintenance team now has to integrate the unscheduled repairs into their already busy maintenance schedule. It can take between days to weeks until the failure mechanism is understood; once that is done, resources must be allocated to do the actual repairs: spare parts must be bought or taken from the inventory and moved to the wellsite, the repair team must be scheduled to go to the site, the well must be made safe to work on, etc.
Depending on the importance of the asset, this process can take weeks to months before any repairs have started. One of the key aims of predictive maintenance is to reduce this time period from weeks to days.
Lower Inventory Costs
If equipment failures can be predicted, the inventory size can be reduced as better planning is possible. Expensive spares can be bought at the right time and stored for smaller time periods. The amount of spares that expire over time without ever being used can be reduced. Also, the number of times critical spare parts might not be available and must be bought at a premium price can be decreased.
Predictive Maintenance can bring large cost savings and extend the life of an oilfield. However, it is not a silver bullet. It might not always make sense to implement it, as it requires additional investments and, maybe more importantly, changes in the organization and changes in established work flows that take time and commitment to execute successfully.
In the next few articles, I will investigate further about the advantages and disadvantages of this technique by building models that simulate the impact of implementation on oilfield production, revenues and costs.