REQUEST PREDICTION

Field of the invention

The invention relates to the field of data analysis and cloud computing in the petroleum industry. More specifically, the present invention is related to digital information handling, storage and simulation associated with various digital rock workflows as may be used, for example, for petrophysical and multiphase flow property evaluation, reservoir characterization, hydrocarbon production analysis, etc.

Background of the invention

There is known a method including receiving, via a network interface of a cloud-based infrastructure, a request from a user for simulating fluid flow in material based on a digital, image-based model of the material (WO 2017079178). Responsive to the request, the simulating is executed via provisioning of one or more resources of the cloud-based infrastructure to generate simulation results; based at least in part on the simulation results the information is transmitted. The cloud-based infrastructure allows to dynamically allocate resources for the computation and provides horizontal scalability of the requests processing. However, the time of doing single simulation is not reduced. The abovementioned method doesn't imply proactive simulation running, so the request from the user is processed as soon as it is generated. This can lead to non- balanced load of the cloud and to high latencies. Summary

The disclosed method comprises collecting requests from a plurality of users received via a network interface of a cloud-based digital rock simulation and database system. Based on the collected requests new user-generated requests are predicted and the cloud-based digital rock simulation and database system is prepared for incoming new user-generated requests in advance.

The new user-generated requests can be predicted using algorithms, written by developers of a target service, or Linear regression, or Bayesian methods, or Decision tree methods, or Neural Networks.

Via the cloud-based digital rock simulation and database system, various data can be accessed: rock data, proppant data or rock data and proppant data, fluid data, chemical data.

The disclosed system comprises servers wherein each of the servers comprises at least one processor, memory accessible by the at least one processor and processor-executable instructions stored in the memory to make predictions of the incoming new user-generated requests, based on the collected requests from a plurality of users. The system also comprises a network interconnect wherein the servers are operatively coupled to the network interconnect, provisioning circuitry that provisions the servers responsive to receipt of a request, transmission circuitry that transmits information based at least in part on simulation results, and a prediction module for predicting the new user-generated requests.

Brief description of the drawings

Fig.1 shows a functional diagram of the Digital Rock cloud system;

Fig.2 shows an example diagram of request processing; Fig.3 illustrates an example of working with the prediction service.

Detailed description

The following description is not to be taken in a limiting sense, but rather is made merely for the purpose of describing the general principles of the implementations. The scope of the described implementations should be ascertained with reference to the issued claims.

This invention describes a cloud-based digital rock simulation and database system (Digital Rock cloud system) equipped with prediction service. In a cloud-based environment, such a system can be accessible by a plurality of different enterprises and users. A cloud-based architecture can be adaptable in real-time within a hardware infrastructure to receive a request for digital data processing, execute a massive pore-scale simulation in response to the request (e.g., from a particular one of a plurality of enterprises), store and handle information on digital rock and fluid data, simulation results and executed scenario(s), and report processed data that can include time-dependent multiparameter three-dimensional graphical output related to one or more performed simulations and stored information.

The difference between the known and the described architecture is in introduction of separate universal microservice, which collects the requests from the user and from the other services, then estimates prediction and suggests to the others, which requests are likely to be in the nearest future.

As an example, a cloud-based system can include various features for physical and digital rock and fluid analyses, which may aid in creating a reservoir model for simulation of flow performance under multiple production scenarios. Such a system may utilize physical laboratory measurements to refine reservoir simulation, for example, to enhance determinations as to relative permeability, capillary pressure, net present value, and other parameters associated with reservoir engineering. As an example, fluid can include liquid and/or gas.

Cloud computing services can be provided via information technology (IT) instruments and technologies that are made available to users in an on-demand manner via the Internet. Cloud computing can allow companies to consume compute resources as a utility rather than having to build and maintain computing infrastructures in-house. Cloud services can provide relatively easy, flexible and scalable access to computing applications, resources and services, and may be managed by a cloud services provider.

An example of a cloud services provider is Amazon Web Services (AWS, Amazon.com, Seattle, Washington), which offers a suite of cloud-computing services that make up an on-demand computing platform. AWS services operate from over a dozen geographical regions across the world. AWS services include Amazon Elastic Compute Cloud, also known as "EC2", and Amazon Simple Storage Service, also known as "S3". AWS services include compute, storage, networking, database, analytics, application services, deployment, management, mobile, developer tools and tools for the Internet of things. AWS services can provide large computing capacity as an alternative to a user having to build a physical server farm.

As an example, a cloud computing platform can be utilized to implement a cloud-based system. For example, consider the AZURE™ platform (Microsoft Corporation, Redmond, Washington), which is a cloud computing platform and infrastructure for building, deploying, and managing applications and services through a global network of data centers.

A cloud computing platform can offer, for example, virtual machines, infrastructure as a service (IaaS) that provide for launch of virtual machines and/or preconfigured machine images, App services, a platform as a service (PaaS) environment (e.g., to publish and/or manage Web sites), Websites, high density hosting of websites (e.g., optionally using one or more of ASP.NET, PHP, Node.js, Python, etc.), etc. As an example, a cloud-based system may utilize Websites in PHP, ASP.NET, Node.js, Python, or one or more other languages. As an example, a cloud computing platform may offer WebJobs as applications that can be deployed to a Web App to implement background processing. Such an approach may be invoked on a schedule, on-demand and/or run continuously. As an example, a cloud computing platform may offer blob (data storage/structure), table and queue services, which may be utilized to communicate between Web Apps and WebJobs and, for example, to provide state information.

A cloud computing platform can provide one or more of SaaS, PaaS and IaaS services and, for example, supports different programming languages, tools and frameworks.

Cloud computing can allow users to benefit from various computing technologies, optionally without deep knowledge about or expertise with each one of them. Cloud computing can reduce, manage and/or control costs. Implementation in a cloud environment can help a service provider to focus on business instead of being impeded by IT obstacles.

As mentioned, cloud services can dynamically scale, for example, to meet demands of users. Provisioning may be automated in a cloud environment where a cloud infrastructure provider supplies hardware and software.

Could computing can be defined in part via the following three application categories: infrastructure as a service (IaaS), platform as a service (PaaS) and software as service (SaaS). As to SaaS, digital rock simulation services and associated digital rock database services can be exposed via the Internet, which can allow customers to use simulation technology as a service in the cloud. Such an approach can mitigate customer costs and allow for on-demand simulation to enhance their productivity as to one or more phases of operations associated with one or more reservoirs.

The cloud-based system can provide digital rock simulation services and associated database services. As an example, one or more reservoir engineers can access one or more of such services via a Web portal or portals to help address current challenges in the petrophysics and reservoir engineering. As an example, a core analyst or core analysts can access one or more of such services to help understand and realistically model pore geometries and fluid behaviors at pores scales in a timesaving manner.

As an example, a cloud computing system can include components for implementing digital rock analysis that integrates physical and digital core techniques, for example, using one or more common rock samples for both types of analysis. Using such a service, oil and gas operators can shorten traditional cycle times, understand better one or more reservoirs prior to making one or more field decisions, and maximize short-term production and long-term recover from oil and gas assets worldwide.

A cloud-based system for digital rock simulation and associated data handling services can address increasing complexity of reservoir formation, fluid behavior and recovery methods. Such a system can provide digital rock simulations in a practically feasible spatial domain and time scale. A cloud-based system can be accessible via the Internet according to appropriate entity accounts (e.g., entity log-in and access credentials).

As an example, a cloud-based system can include digital rock models for generating simulation results via one or more simulators (e.g., which may be instances of simulators in a cloud environment). In such an example, simulation results can include a substantial amount of numerical information and associated data that can be properly and securely accumulated and stored, for example, to provide for analysis, archiving and/or retrieval.

A cloud-based system can include a process architecture and corresponding infrastructure for digital rock and fluid data handling, storing and efficient numerical simulations supported by cloud-based high performance computing technology with a capability for remote access and control via the Internet.

A cloud-based system can be accessible, interactively, via networked client devices, such as workstation, desktop computers, laptops, tablets and mobile phones. Customer access can also be applied for remote visualization of 2D and 3D initial, processed and simulated images where image rendering may be performed, at least in part, by a cloud server with dedicated software and hardware. As an example, one or more cloud-based servers can receive one or more viewpoint requests from a device of a customer and, in response, transmit one or more corresponding 2D or 3D rendered images back to the device of the customer.

Fig.1 shows an example of functional diagram of the Digital Rock cloud system. Every operation starts with sequence of the requests from the user 101, which are generated by the user interface 102 and come to the cloud 103 via cloud gateway 104. Then the system configures the processing modules 105, which can be data processors, visualization modules, simulators, etc. The processing modules can utilize storage 107 and database 109. When the processing module is finished, it generates a response which is returned to the user.

Besides this way, the system can work with the prediction service 108. The requests, generated by the user, and by the other parts of the cloud, are coming also to the prediction service. The prediction service has a set of rules, with configured predictions. These predictions are requests, or their parts, which are anticipated in the future. Every rule has destination module which is informed of incoming prediction. Since the predicted request is come to the target module, it can be processed in the same way as an actual request, described in the previous example.

The cloud-based digital rock simulation and database services system can include components for remote access which will help a customer to connect and interact with the system from various interfaces including web-based interfaces, remote desktops and remote client applications.

Fig.2 shows an example diagram of request processing. The gateway 110, which received a request 112 from the user, performs consistency check 113, and creates executor instance 114. The pipeline executor module 111 consists of the following steps: the pipeline configuration is requested from the database 116, pipeline assembly 117, input data download 118, pipeline running 119, output data upload 120, report generation 121. Then the report is handled to the user 115.

The cloud-based system can include visualization services that include, for example, features to perform 3D visualization of digital rock input data images, digital fluid data and simulation result representation including the comparative and sensitivity analysis of various simulation scenario runs. As an example, 3D visualization can be utilized to control acceptability of a simulation along computational time, for example, to highlight volumetric properties that may not be readily amendable to expression numerically and, for example, to analyze and record time-dependent 2D or 3D graphical output of one or more performed simulations.

The prediction module is considered as a separate service inside the cloud system. Generally, the behavior of the service is as follows:

• Listen to the requests from other services and from the user and collect them. • Predict the future requests

• Send them to the target services

• Perform training iterations in background.

An example of the prediction service workflow is illustrated in Fig.3. Once the request came inside the prediction service 122, it is enriched with contextual information 123. Then, according to the defined rule, the prediction service applies importance metric of the input information 124. Then the prediction is performed 125, and sent to the target service. After that, when the prediction continues to collect requests 126, which can be used for prediction training 127.

The result of the prediction is a set of requests, which are likely to be in the nearest future. If the prediction method allows to estimate probabilities and confidence of prediction, this information is also accessible. The prediction methods are not fixed and can be chosen. The list of the methods can include:

1. Algorithmic rules, defined by the developer of the target service

2. Linear regression

3. Bayesian methods (Naive Bayes, Bayesian network, etc.)

4. Decision tree methods

5. Neural Networks

The training model can be chosen from two options: statistical and interactive.

1. Statistical mode. The requests are collected and recorded. The system is being trained eventually, e.g. when the cloud is under-loaded.

2. Interactive mode. The system is being trained while the requests come. This mode can be considered as reinforcement learning problem. Working in this mode is helpful for predicting the human interaction, e.g. for user experience improvement.

Some embodiments of the disclosed method are described below. 1. Data prefetching

Some routine procedures in Digital Rock cloud system require to load considerable amount of data before starting operation. The prediction of the future data requests allows to pre-load this data and to reduce the loading latency.

2. Operations preparing

If some service require time to prepare for running, it can be done as soon as it is predicted. Thus, the latency of operation is reduced.

3. Operations pre-running

Depending on the confidence level, cost of the operation, and its priority, some operations can be precomputed. The risk of doing useless job is balanced with the profit of immediate actions.

4. Proactive cloud scaling

It is common for the cloud architecture to scale the resources when the load is high. The prediction of the load level allows to scale the cloud proactively, therefore improves stability of the cloud applications.

5. User Interface improvement

The most likely request from the user side, if predicted, can help to customize user interface. The following items can be introduced:

• The most probable parameters can be filled as default options.

• If the options are represented as a list, the most probable items can be highlighted or moved to the top, as well as the least probable items can be dimmed or hidden

• Pipelines can be identified and suggested to the client (or even precomputed) for similar datasets.

6. Attenuation of Internet lag in interactive applications Some interactive applications, like 3D visualization, suffer from Internet lag. The request prediction can make user interaction smoother and will increase client satisfaction.