Learn about data reconciliation and how we apply the concept to provide our customers with the best possible flow rate estimates given all their measurements.

Single best estimate
Combine flow rate information from different sources into a single source of truth
Utilize topology
Estimate flow rates for wells with no direct flow rate measurements
Decrease uncertainty
Decrease uncertainty of measured flow rates

The challenge of combining all available flow rate information
Many oil and gas production systems have varying degrees of redundancy for different wells. Some wells have several redundant rate sources, with multiphase flow meters (MPFM) and virtual flow meters (VFM) ensuring real-time rate monitoring. Other wells have their flow rates measured only intermittently, e.g. when they are routed to a test separator. In some cases, there might even be wells without a direct way of measuring their flow rates. For example when the produced hydrocarbons are mixed with production from other wells before this commingled flow is measured.

There are numerous possible setups, but there exists a method that can be applied to all kinds of topologies and instrumentation setups to provide the most accurate flow rates.

Imagine a production system like the one in the illustration. A is a well with an MPFM, and B is a well with no direct measurements. The flow from wells A and B are commingled before the combined flow is measured first by an MPFM in C and then a separator in D. Assuming steady-state production, we know by the law of mass balance that the flow through MPFM C and separator D must be the same. Knowing this, we can calculate the most likely flow through MPFM C and separator D as a weighted average of the two measurements, depending on the uncertainties of the MPFM and the separator. Also, from the mass balance, we know that the sum of the flow from well A and well B must be equal to the commingled flow. We can now calculate the estimated flow rate of well B as the commingled flow minus the MPFM measurement of well A.


A data-driven approach to smarter allocation and gross error detection

FlowFusion (FF) is our industry-tested software service for reconciliation and allocation. It is a fully data-driven approach that exploits the information that lies in production data, such as quantifiable uncertainties, well tests, redundant flow rate data, and detectable MPFM or VFM errors. The main methodology used for reconciliation is data validation and reconciliation (DVR), yet FF is also much more than just the reconciliation method.

We generally divide FF into four modules as illustrated below: data processing, uncertainty estimation, the reconciliation problem, and gross error detection.

Data preprocessing
The first module is about data processing and is essential to tackle poor data quality and ensure realistic and accurate rate estimates.
The second module involves estimating the uncertainty of the different flow rate signals that are used in the problem. The uncertainties are rarely known exactly and are usually be guesstimated together with production engineers that know the wells.
The third module is the reconciliation, which solves the DVR problem for whichever desired time period and production system setup given the validated data from the “data processing” step and the variances calculated in the “uncertainty estimation” phase.
Error Detection
The last module, gross error detection (GED), can be used to warn about abnormalities related to any of the sensors, for instance, drifting sensors.

Outcome - one high-quality soft tag for each flow phase

Flow rate reconciliation provides value for production teams, both for small and well-instrumented assets or large assets with fewer sensors per well. For a well with redundant measurements, it can help the engineer combine the measurements into one single source of truth for each of the phases. For sparsely instrumented systems the value is even greater, where this approach can estimate flow rates in parts of the system where there is little or no (real-time) knowledge of the flow rates.

FlowFusion as a service

As with NeuralCompass VFM, FF is offered as-a-service. We connect automatically to your databases and if set up, FF is smoothly integrated with the Well Test Application. FF therefore operates with low intrusiveness in your everyday work life and can, if you allow it, be your new colleague for smarter allocation.

Error reduction: 38% for oil, 6% for gas, and 49% for water

Using FlowFusion instead of a standard industry approach

FlowFusion Highlights

Delivers the most likely flow rate estimates for your wells by applying data validation and reconciliation, mass balance enforcement, and live, condition-based uncertainties

Low intrisiveness

FF is operated though an automated workflow, fetching and writing back data in a way that causes little intrusiveness in your everyday routine

Handy error detection

Using statistical methods, FF notifies you about faulty or deviating rate measurements

Single source of true

Fuses all available rate information together to give you robust flow rate estimates that reduce biases and variance

Flexible Framework

A highly flexible framework that can handle a variable number of available measurements and be tined to your liking

Real-life results
The figure below illustrates an example where the data processing module comes in handy. Here we see that VFM 1 has suddenly dropped to a new and likely faulty value. With two algorithms for error detection, stray and frozen detections, this behavior can be easily captured and, given the sensor redundancy, such sensors can be automatically filtered out of the reconciliation problem. Observe, for instance, that the white FF rate is not drawn towards the faulty estimates of VFM 1 - evidence that VFM 1 is filtered out.

You can find more information about FlowFusion in one of our newest publications:

Flow Fusion, Exploiting Measurement Redundancy for Smarter Allocation