Providing customers with live data-driven flow rate predictions
The model is trained. It performs excellent on the test data. That means that we are finished with our work, right? Actually, the answer is no. Designing and training the model is just half the story.
![Author Kristoffer Nesland portrait image](https://solutionseeker2021.imgix.net/images/people/kristoffer_avatar.png?auto=compress%2Cformat&crop=focalpoint&cs=srgb&fit=crop&fp-x=0.5&fp-y=0.5&h=160&q=90&w=160&s=cf830df2f9d7c2876ba22d5d3045090c)
- Author
- Kristoffer Nesland
- Publish date
- · 3 min read
The real value for the customer is not unleashed before they receive live flow rate predictions on how much oil, gas and water each of their wells produce. To accomplish this, we need to put our multi-task learning model into production.
Four steps are required to provide the customers with live flow rate predictions for a well. First, we stream temperature, pressure, valve opening and potentially other sensor data from the well. Then, we resample the sensor data to make sure that all of the time series are aligned and have values at the same points in time. Now, we are ready to feed the model with the data. Every time a set of new sensor data arrives from the well, the values are passed through the neural network together with information on which well the values originate from. After passing through all the hidden layers of the model, the input is transformed into three floating point numbers, one for each of oil, gas and water, all in standard cubic feet per day. Finally, the newly generated predictions are streamed back to the customer and find their way into various other use-cases, like allocation, production optimization and reservoir model update.
![](https://solutionseeker2021.imgix.net/images/2.5-mtl_serving.png?auto=compress%2Cformat&crop=focalpoint&cs=srgb&fit=crop&fp-x=0.5&fp-y=0.5&h=338&q=90&w=600&s=b16d4fe370e14a452b1f448031c01352 600w, https://solutionseeker2021.imgix.net/images/2.5-mtl_serving.png?auto=compress%2Cformat&crop=focalpoint&cs=srgb&fit=crop&fp-x=0.5&fp-y=0.5&h=394&q=90&w=700&s=ccddcb6ff096033b030054224b01082f 700w, https://solutionseeker2021.imgix.net/images/2.5-mtl_serving.png?auto=compress%2Cformat&crop=focalpoint&cs=srgb&fit=crop&fp-x=0.5&fp-y=0.5&h=451&q=90&w=800&s=0322b85c57ed27b0b3d35df63815de9c 800w, https://solutionseeker2021.imgix.net/images/2.5-mtl_serving.png?auto=compress%2Cformat&crop=focalpoint&cs=srgb&fit=crop&fp-x=0.5&fp-y=0.5&h=507&q=90&w=900&s=398fc9286c390b1b915c34436168898f 900w, https://solutionseeker2021.imgix.net/images/2.5-mtl_serving.png?auto=compress%2Cformat&crop=focalpoint&cs=srgb&fit=crop&fp-x=0.5&fp-y=0.5&h=563&q=90&w=1000&s=f004ae50ad944c14cbeb21be3fd0aa4e 1000w, https://solutionseeker2021.imgix.net/images/2.5-mtl_serving.png?auto=compress%2Cformat&crop=focalpoint&cs=srgb&fit=crop&fp-x=0.5&fp-y=0.5&h=620&q=90&w=1100&s=8247625efd95ac2a2092438da67f30cb 1100w, https://solutionseeker2021.imgix.net/images/2.5-mtl_serving.png?auto=compress%2Cformat&crop=focalpoint&cs=srgb&fit=crop&fp-x=0.5&fp-y=0.5&h=676&q=90&w=1200&s=a83a1925bd7f7b685a4b158021d3b3e0 1200w, https://solutionseeker2021.imgix.net/images/2.5-mtl_serving.png?auto=compress%2Cformat&crop=focalpoint&cs=srgb&fit=crop&fp-x=0.5&fp-y=0.5&h=732&q=90&w=1300&s=9107211ea768b8dc6bb0d0f715c04b72 1300w, https://solutionseeker2021.imgix.net/images/2.5-mtl_serving.png?auto=compress%2Cformat&crop=focalpoint&cs=srgb&fit=crop&fp-x=0.5&fp-y=0.5&h=789&q=90&w=1400&s=14fc40c452c06896ef3a409c6f8446c4 1400w, https://solutionseeker2021.imgix.net/images/2.5-mtl_serving.png?auto=compress%2Cformat&crop=focalpoint&cs=srgb&fit=crop&fp-x=0.5&fp-y=0.5&h=845&q=90&w=1500&s=61c80865e565975150511ba81e6c7970 1500w, https://solutionseeker2021.imgix.net/images/2.5-mtl_serving.png?auto=compress%2Cformat&crop=focalpoint&cs=srgb&fit=crop&fp-x=0.5&fp-y=0.5&h=901&q=90&w=1600&s=07a7b008cff5bf7efc3adae85692116e 1600w, https://solutionseeker2021.imgix.net/images/2.5-mtl_serving.png?auto=compress%2Cformat&crop=focalpoint&cs=srgb&fit=crop&fp-x=0.5&fp-y=0.5&h=958&q=90&w=1700&s=d4234c7a0cb0096de4b6663dc9738a08 1700w, https://solutionseeker2021.imgix.net/images/2.5-mtl_serving.png?auto=compress%2Cformat&crop=focalpoint&cs=srgb&fit=crop&fp-x=0.5&fp-y=0.5&h=1014&q=90&w=1800&s=6af724faca615548d2817ec3c6987c42 1800w, https://solutionseeker2021.imgix.net/images/2.5-mtl_serving.png?auto=compress%2Cformat&crop=focalpoint&cs=srgb&fit=crop&fp-x=0.5&fp-y=0.5&h=1070&q=90&w=1900&s=6777303b876460877969c188e7929312 1900w, https://solutionseeker2021.imgix.net/images/2.5-mtl_serving.png?auto=compress%2Cformat&crop=focalpoint&cs=srgb&fit=crop&fp-x=0.5&fp-y=0.5&h=1127&q=90&w=2000&s=6a0f1a22568d5accfbc2e96a97a405d5 2000w, https://solutionseeker2021.imgix.net/images/2.5-mtl_serving.png?auto=compress%2Cformat&crop=focalpoint&cs=srgb&fit=crop&fp-x=0.5&fp-y=0.5&h=1183&q=90&w=2100&s=4872c994c40c78ff3721389fbed04d2b 2100w, https://solutionseeker2021.imgix.net/images/2.5-mtl_serving.png?auto=compress%2Cformat&crop=focalpoint&cs=srgb&fit=crop&fp-x=0.5&fp-y=0.5&h=1239&q=90&w=2200&s=0c007e39f3dd1430eb505cf3c831d155 2200w, https://solutionseeker2021.imgix.net/images/2.5-mtl_serving.png?auto=compress%2Cformat&crop=focalpoint&cs=srgb&fit=crop&fp-x=0.5&fp-y=0.5&h=1296&q=90&w=2300&s=ba5a39860e9da9c0c26c3b2abcbc4143 2300w, https://solutionseeker2021.imgix.net/images/2.5-mtl_serving.png?auto=compress%2Cformat&crop=focalpoint&cs=srgb&fit=crop&fp-x=0.5&fp-y=0.5&h=1352&q=90&w=2400&s=0a7e5a4ad5f476601925b2733399024e 2400w)
As seen in the previous posts about physicality and accuracy, the multi-task architecture provides benefits when it comes to prediction accuracy and robustness over time. Coincidentally, it turns out that the multi-task architecture also has some benefits when it comes to serving the rates live. To understand why, imagine that you would like to provide flow rate predictions for 50 individual wells. Without the multi-task architecture, you would need to train and keep track of 50 individual machine learning models. When time passes and new well tests are recorded you would need to re-train several models. Keeping all of these models up to data can become time consuming compared to a single model for all of the wells. A 50 well multi-task model will typically need significantly fewer parameters than 50 individual models combined. An added benefit here is reduced time needed for training the models. In the case of 50 wells, we have seen a reduced training time by a factor of around 4. All in all, improvements like this contribute to less time spent baby-sitting the models and higher uptime.
In Solution Seeker, we have a strong focus on productifying our research. The work on multi-task learning for flow rate estimation presented here is a prime example of this. As seen, the approach leads to solid research results and at the same time it simplifies the process of providing our customers with flow rate estimations. Solutions like this prove that high performance does not always need to come at the expense of additional complexity.
If you would like a demo or discuss data-driven flow rate solutions for your asset, don’t hesitate to get in touch.