Identifying Correct Oil-Well Bean Up Profile

Situation

‍

At this plant, oil wells have a choke to control the flow from the well. This choke both decreases the chance of instabilities and helps to control the pressure. As a result, blow-outs can be prevented which is important to operate at maximum efficiency. When wanting to startup a well after a shutdown, the plant needs to get oil flowing from the reservoir, and the flow rate needs to be increased in a controlled manner. The profile that leads towards the operating flowrate is called the bean-up profile. If this profile is not satisfied, instabilities will cause a well to shutdown fast. Therefore, to avoid this issue, an optimal bean-up profile must be identified to more accurately monitor the well start-up behavior and prevent fast shutdowns.

Problem

‍

Process experts needed to determine the proper bean-up profile. To do this, they needed to search the historical data for good bean-up profiles in order to create a fingerprint to use as a comparison in the future. However, identifying the correct bean-up profiles is difficult. Traditionally, fingerprinting a golden bean-up profile required data science and coding expertise which is often time-consuming and expensive

‍

‍

Solution

‍

Process experts used TrendMiner to easily and quickly search historical data to identify good startup periods. They created a fingerprint that they could use as a basis and set a monitor to alert personnel for startup deviations.

‍

‍

Approach

‍

• The process experts searched into the historical data and overlay different timeframes which included loading the choke state and the flow tag.
• They next searched the historical data to find both bad and good startup profiles and defined a good bean-up profile. They also searched for the startup periods to determine the time it took for these startups.
• The team discovered that a too fast bean-up led to a low
  time of operation, so used filters to get rid of the bad runs.

‍

Results

‍

• Historical data was used to identify the relationship between a bad oil-well run and a too fast bean-up.
• Historical data was used to overlay good start up profiles to create a golden fingerprint of this startup profile and used it as a comparison for all start up profiles.
• A monitor was set up to send alerts to personnel in case of a too fast bean-up process, giving them sufficient time to proactively take corrective measures.

‍