Despite the advent of ever more sophisticated virtual flow metering techniques; engineers and modelers will always need to know for sure how much oil, water and gas the well is producing in order to calibrate models and verify the AI technology is working as expected.
Traditionally, well testing relied on the measurement of liquid volumes, rather than flows; and the results of lab data for precise GOR and water cut measurements; however in the last 20 years flow meters and online analysers have become more and more accurate and almost all well tests rely on separation of the phases and individual phase flow measurements.
The latest field developments would probably chose to install multiphase flow meters (MPFM) which have long promised to do away with large and expensive test separators; but in my mind a test separator and the measurement of the different phases will always be required to calibrate and validate the measurements of the MPFM. I would love to read your thoughts if you think this is no longer the case!
Talking about digitalisation there are three areas where the production well testing process can be optimised as defined by the pioneering PRAP project in the late 90’s:
- Well Test scheduling, where the schedule of tests to be executed is optimised to test first the wells that most likely would have changed their behaviour; or those with an overdue test for legal or internal purposes.
- Well Test execution, where the well test duration is minimised to produce a valid result. Minimising the duration of the test would enable to test more wells in a given span of time.
- Well Test results analysis, where the results of the test are validated using AI techniques and by comparison to first-principles models and/or aid is given to the engineers in charge of validating the results.
The relative importance of the first two will depend on the number of wells per well testing facilities. The early PRAP project, which was developed for facilities where there was a ratio of about 30 wells per test separator, developed rules for the preparation of an optimal well test schedule and an algorithm to detect convergence of well test results that managed to reduce the well test duration by 50%; however, these processes do not seem to be critical for most fields in operation today.
Regarding the well test analysis process, a series of heuristic rules were developed to automatically reject any obvious bad tests and automatically request a re-test. This is precisely the process that has seen several incarnations over the years in the form of tools and applications that facilitate and optimise the well test analysis process.
Back in 2006 and 2007, Eigen developed and deployed a well test analysis portal for an operator in the Caspian Sea that is still in use today: The portal was developed using ASP with an IP21 stored-procedure backend and would allow the engineer to select and re-calculate the stable well testing period, to compare the results with a coefficients-based well production model and to look back at the previous results for the same well. At the end of the process, the user can accept or reject the well test results.
A good well test analysis tool should allow the users to:
- Automatically capture and present well tests that have not been analysed
- Automatically or manually select the stable measurement period
- Using AI techniques and even simple heuristic rules, automatically select and pre-classify the well test results, pending engineering approval
- Integrate and compare with first-principle models, even allowing the update of the models or detecting the need to re-test
- Allow graphical comparison with previous well tests for the same well/completion
- Facilitate the process of accepting or rejecting well tests
- Storing the well test results and other metadata in the approved system of record
At Eigen, we have 30+ years of experience developing applications that support and optimise Oil & Gas operations. Check out our products and custom development services that can address your particular engineering and optimisation needs.