What's the best reservoir simulation software for your project?

It’s funny to see people advocating an AI approach that does not require a geological model. Isn’t it every reservoir engineer’s dream to get rid of geology? I guess these folks never got the memo on the virtues of interdisciplinary teamwork. AI without geology and physics? What could go wrong?

Mr. Dan O'Meara has mentioned "AI without geology and physics?" I am sure he has not read my explanation of Artificial Engineering Intelligence (AEI) difference with Artificial General Intelligence. AEI is absolutely "WITH" not Without "geology and physics". Using AI-Reservoir Modeling, Geo-Analytics (AI-Based Reservoir Geological Model) is developed using all the geological-based actual data, and unlike NRS software (CMG, Eclipse, Petrel, tNavigator, ...) it does not need a geological model that reservoir engineers must have to modify it to achieve History Matching and EOR. AEI requires expertise of Engineering and Expertise of AI. When you do not know what AI is, then yes, you are correct, you cannot use it to do anything in Engineering.

I recommend Artificial Intelligence-Based Reservoir Simulation and Modeling software. Numerical reservoir simulators are significant tools to estimate and optimize the performance of oil and gas fields under various scenarios. They are usually obtained by the finite difference or finite element method. Although numerical reservoir simulators are excellent tools for field studies, they have limitations: 1. Simulation requires hours to run a single scenario due to computational complexity. 2. Simulation is impractical to obtain optimal solutions given a limited set of scenarios. 3. Traditional models cannot process the large number of parameters required to increase resolutions. 4. Computationally expensive and time-consuming.

The best reservoir simulation software is the one you understand, i.e., not a black-box. Some are a lot cheaper than others, or quicker, some excel at integration, some excel at physics. However, excellent physics may not always add value. Minimum-viable or good-enough functionality is often the best approach. A lot of features and choices are negotiable, but, IMO, pushing buttons on a black-box is not acceptable.

I fully agree. One aspect that requires a lot of attention is that of uncertainty management. Any model is as good as its input and a full understanding of the uncertainty ranges of the input is essential to judge the quality of the predictions and thereby the quality of the decisions based on these predictions.

The availability of an infrastructure layer must be taken into account first in selecting the best candidate application. Building on an infrastructure that is already in place will be the best option, while the need to introduce an appropriate and reliable infrastructure will dictate choosing the right application which, also, need to be reliable and economical. It is also important to consider the scalability of the software. Can it handle large-scale simulations and complex models? how easy is it to extend it to include more projects or even adding extra components in the current project?

Eclipse simulator, from SLB, is the reservoir simulator with the best integration with geological models, well logs and production data, because is interconnected with Petrel. The user environment for the Eclipse is Petrel RE, which integrates the static and dynamic modeling process into a seamless workflow.

Petrel software & it enables the user to interpret seismic data, perform well correlation, build reservoir models, visualise reservoir simulation results, calc volumes, produce maps and design development strategies to maximize reservoir exploration

Estoy totalmente de acuerdo , el simulador debe ser amigable , y que permita tener un lenguaje que permita la carga rápida de datos . Que pueda tener respuestas dependiendo el nivel de detalle que se necesite, respuestas rápidas y otras más elaboradas .

Coming from a software company background where I was mostly involved in software training and support for a few years, I have come to appreciate the need for software training post acquisition. Most software sales and contracts do include training because the software is of no use without proper training to begin with. But coming from academia where I currently find myself, some software are donated to schools and no training is added which makes it difficult to use. One would incline to a particular software based on user friendliness of the GUI or ease of navigating the software if the user guide is easy to follow. Training and access to support is very important factor to consider in software acquisition.

One of the best ways to fill in the knowledge gap is to always read the new release notes and updates from the vendors. Most of the vendors release comprehensive updates and new documents that explain additional keywords and how to optimise both the performance and speed of the simulators.

To prevent any misunderstandings, it's crucial to include the support details in the contract and make sure the terms and expectations are clearly stated. The agreement should include details such as the hours of support availability, response times for resolving issues or answering inquiries, and the methods of communication (e.g., phone, email, ticketing system). It is crucial to state weather the training is included in the initial contract or if it incurs additional charges for certain levels or types of training.

I agree. REs should always ensure the keywords been twea?ed in their models to obtain quick results are reasonable and with fair justifications.

Petrel Reservoir Engineering: Petrel, developed by Schlumberger, is an integrated software platform that includes reservoir simulation capabilities. It is known for its ease of use and integration with other geoscience and engineering tools.

t-Navigator, however: When considering a reservoir simulation software, cost and value are crucial factors to evaluate. It is important to assess the expenses involved in purchasing, installing, licensing, updating, and maintaining the software. Additionally, compare it to alternative options in terms of features, performance, quality, and reputation. Ensure it aligns with your budget and return on investment expectations, while helping you achieve your project goals. Prioritize software that offers the best value for your money, but also consider factors like functionality, user interface, and technical support. Make an informed decision based on a comprehensive evaluation to select the most suitable software for your project.

Yes absolutely, Petrel software licensing costs can vary widely based on your specific needs and usage, ranging from thousands to hundreds of thousands of dollars per annum, and onetime use or short-term licenses may provide cost-effective options for specific projects.

Cost of the simulator can be small compared to the capital decisions, but large compared to the cost of a person. So software cost and value should be focused on reducing heterogeneity in the organization to achieve economy of scale on people training and succession planning, and adapting to surges in demand with a SaaS style implementation. Also, very fundamental, any history match project should have a clear business goal - a new well, an upgrade to a facility, or production pattern balance. Different projects will have different requirements for signal to noise in the predictions, and different areas of interest. If the cost of the forecast is less than 1% of the capex of the project, you may want to check what you're missing.

I do agree with this. My experience took me to work with some clients in evaluating my former company's reservoir simulation software but the client chose to buy a cheaper one that offered more licenses for its engineers compared to mine which was one user per license. So good price with high value is of much interest when buying any reservoir simulation software.

About CMG, Complexity of Models: CMG relies on mathematical models to simulate fluid flow in subsurface reservoirs. These models are simplifications of complex geological and fluid properties. The limitations become apparent when the reservoir has highly heterogeneous rock properties that are difficult to model accurately. For example, if a reservoir contains intricate fracture networks, modeling these networks accurately can be challenging.

WEM (Well Evaluation Model) by P.E. Moseley and Associates is used to design wells for optimum production, monitor performance, and troubleshoot and resolve problems. WEM engineering technology includes completions, production, artificial lift, well testing and phase behavior to facilitate quick, accurate solutions for virtually any well performance issue.

"Review and feedback" title is missing another important term "implementation time". Every provider has user meetings, etc. But not everyone is good at integrating the feedback. I hope "big boys" have changed their ways, but in the past, when I was using Eclipse, we were getting confirmation that certain fixes and improvements requested by us and other multiple users were in the pipeline. Actual implementations were observed many years later. With smaller new players we had amazing implementation time. Things might slow down a bit with increased user base, but still really good.

Reviews are important as they actually help the software companies to work on these software so they continue to be competitive on the market. R&D divisions of these reservoir simulation software companies follow through feedbacks and reviews from their clients to improve on the capabilities of the software so its another imperishable factor to consider in purchasing a software.

Show me your reservoir initialization and I’ll show you how it fails to honor both physics and basic input data. Here is a quick test. Try cross plotting the 3D initial water saturation versus the pore volume weighted water saturation log. With rare exceptions, you’ll encounter a mess. When your starting point bears so little connection to physics and data, is it any wonder it takes months to get a match? If you wish to reply to my comments, please include this cross plot for your model.

Which software's used doesn't matter. It's just a tool. Most sim models didn't need to be built in the first place. Anytical modelling and production analysis as pre-requisite to a sim model is what's lacking!

I always want to emphasize a"fit-for-purpose" reservoir simulation strategy. It's essential to frame the project's needs first, such as reservoir type, size, complexity, and available resources, and consider factors like technical support, training, and future scalability. Sometimes a tank model or a type curve is good enough. Sometimes a detailed full-field earth-model based simulation is the best option. Based on the project needs, we could evaluate the available options and pick the right software.

The best simulation software is the one which is suitable for scenarios one’s want to test. The software is a tool , in other word an assistant the key player is the interpreter who knows want he wants to achieve. Of course there are cases where technical limitations of the software are what force to select one instead of another. When it comes to how a software is user friendly, this topic is very subjective because what is considered as user friendly for one user could be the opposite for another user. The most important is making sure that the software offers all the functionality the user wants in order to run his simulation cases

Critical consideration should be given to nontechnical issues and forecast assumptions should incorporate these risks. For example, gas demand for power generation might reduce during the raining season (in tropical jurisdictions) when hydro power becomes a preferential power source for various reasons (GHG emissions reduction, OPEX reduction etc.). The ripple effect is that wells will need to be beaned down to reduce AG production. All this links back to the subsurface (e.g reduced pressure drawdown etc). A probabilistic approach to forecasting is important in defining the true base case. A lot of the times a deterministic approach will likely gravitate towards the optimistic (high case), so seeing the band gives a realistic perspective.