Monitoring of Wellsite with Artificial Intelligence

Control and/or monitoring of equipment on remote oil and gas wellsites can be challenging. Oftentimes, the wellsites remain unoccupied by engineers and technicians, and are only occasionally visited for routine maintenance and/or for offloading gas and/or oil. Thus, levels of storage tanks and the functioning of equipment is oftentimes either left unmonitored or is monitored by rudimentary sensors, that are only able to obtain basic information. Furthermore, the output of those sensors generally needs to be interpreted by an individual, which requires additional manpower, introducing inefficiencies in the extraction of oil and gas. Embodiments of the present disclosure can reduce many of the inefficiencies associated with an oil and gas wellsite and provide for more automated control over the wellsite.

Various embodiments of the present disclosure include a method for monitoring a wellsite. In some embodiments, the method can include receiving data from a wellsite sensor. In some embodiments, the method can include analyzing the data using artificial intelligence to determine a characteristic associated with the wellsite.

Various embodiments of the present disclosure include a system for monitoring a wellsite. In some embodiments, the system can include a processor. In some embodiments, the system can include a memory storing non-transitory computer-readable instructions that are executable by the processor. In some embodiments, the computer-readable instructions can be executable by the processor to receive data from a wellsite sensor, wherein the wellsite sensor includes a camera and the data includes image data. In some embodiments, the computer-readable instructions can be executable by the processor to analyze the data using computer vision to determine a characteristic associated with the wellsite. In some embodiments, the computer-readable instructions can be executable by the processor to record data associated with the characteristic in a data log.

Various embodiments of the present disclosure include a system for monitoring a wellsite. In some embodiments, the system can include a processor. In some embodiments, the system can include a memory storing non-transitory computer-readable instructions that are executable by the processor. In some embodiments, the computer-readable instructions can be executable by the processor to receive image data from the camera. In some embodiments, the computer-readable instructions can be executable by the processor to analyze the image data using computer vision to determine a characteristic associated with the wellsite. In some embodiments, the computer-readable instructions can be executable by the processor to determine, with artificial intelligence, a type of notification to generate, based on the characteristic associated with the wellsite.

Embodiments of the present disclosure are described below with reference to the accompanying figures. The features and advantages which are explained are illustrated by way of example and not by way of limitation. One of ordinary skill in the art will recognize that there are additional features and advantages provided by embodiments of the present disclosure beyond those described herein.

depicts a systemfor monitoring a wellsite with artificial intelligence, in accordance with embodiments of the present disclosure. As depicted, in, in some embodiments, the wellsite can include one or more of a number of process equipment. In an example, the process equipment can include one or more of a storage tank, a flare, well head, a separator, and/or other equipment not shown. As will be appreciated, the wellsite can include other equipment not shown, such as pumps, valves, distillation columns, pump jacks, etc. In some embodiments, as discussed herein, wellsites can be located in remote regions and may not be regularly manned by personnel. As such, it can be beneficial to have ways to monitor the status of and control day to day operations of the wellsite, which can allow for selective elevation of incidents to personnel and/or to a central control and command center.

Accordingly, embodiments of the present disclosure can utilize a monitoring and/or control system that can utilize artificial intelligence to monitor and/or provide control over the system. As depicted in, the system can include a number of sensors, which can monitor various aspects of the system. In some embodiments, the sensors can include cameras-,-,-, hereinafter referred to in the plural as cameras. In some embodiments, the camerascan include an image sensor and/or a thermal sensor and can be a still camera and/or a video camera. In some embodiments, the camera can include an optical gas imager, which can include a cooled or uncooled microbolometer with an appropriate spectral response range and sensor sensitivity to target specific hydrocarbon gases. As further discussed herein, a camera can include an optical gas imager.

As depicted, the first camera-can be directed towards the storage tank, such that at least a portion of the storage tankis within view of the camera-. As further depicted, a second camera-can be directed towards the flare, such that camera-captures a picture of a plumeof the flare. In some embodiments, a third camera-can be directed towards a capassociated with a wellbore, and/or associated piping-,-, . . . ,-. Although embodiments are discussed in relation to a wellsite, embodiments of the present disclosure can also be used at other types of sites. For example, embodiments of the present disclosure can be used at a transfer station for natural gas, a waste management site and/or landfill, a mine, etc.

With respect to the storage tank, in some embodiments, the whole storage tankmay be within view of the camera-and/or a select portion of the storage tankmay be within view of the camera-. In some embodiments, the first camera-, and/or another camera, depicted incan be directed towards multiple pieces of equipment that are included in the system. For example, in some embodiments, one or more of the camerascan be directed towards one or more pieces of equipment, such that data collected from the cameras(e.g., in an image and/or video) captures one or more pieces of equipment. In some embodiments, a single camera can be used to monitor one or more pieces of equipment. Furthermore, one or more cameras can monitor other aspects associated with the wellsite, such as vehicles, personnel, intruders, etc.

In some embodiments, the system can include a computer. In some embodiments, the computermay be located at the wellsite. In some embodiments, the computermay not be physically located at the wellsite. For example, in some embodiments, the computercan be a cloud based computing system and/or can be located at a remote site, such as a control center. In some embodiments, the computercan include a processorand a memory, which include memory resources (e.g., volatile memory and/or non-volatile memory) for executing instructions stored in a tangible non-transitory medium (e.g., volatile memory, non-volatile memory, and/or machine readable medium) and/or an application specific integrated circuit (ASIC) including logic configured to perform various examples of the present disclosure. A machine (e.g., a computing device) can include and/or receive a tangible non-transitory machine readable medium storing a set of machine readable instructions (MRI) (e.g., software) via an input device.

As used herein, processorcan include one or a plurality of processors such as in a parallel processing system. Memory resourcescan include memory addressable by the processor resourcesfor execution of machine readable instructions. The machine readable medium can include volatile and/or non-volatile memory such as random access memory (RAM), magnetic memory such as a hard disk, floppy disk, and/or tape memory, a solid state drive (SSD), flash memory, phase change memory, etc. In some examples, the non-volatile memory can be a database including a plurality of physical non-volatile memory devices. In various examples, the database can be local to a particular system or remote (e.g., including a plurality of non-volatile memory devices).

The processor resourcescan control the overall operation of the system. The processor resourcescan be connected to a memory controller, which can read and/or write data from and/or to volatile memory (e.g., RAM). The memory controller can include an ASIC and/or a processor with its own memory resources (e.g., volatile and/or non-volatile memory). The volatile memory can include one or a plurality of memory modules (e.g., chips). A basic input-output system (BIOS) for the systemmay be stored in non-volatile memory or other non-volatile memory not specifically illustrated, but associated with the processor resources. The BIOS can control a start-up or boot process and control basic operation of the system.

The processor resourcescan be connected to a bus to provide for communication between the processor resourcesand other portions of the system. For example, the bus may operate under a standard protocol such as a variation of the Peripheral Component Interconnect (PCI) bus standard, or the like. The bus can connect the processor resourcesto the non-volatile memory, graphics controller, input device, and/or the network connection, among other portions of the system. The non-volatile memory (e.g., hard disk, SSD, etc.) can provide persistent data storage for the system. The graphics controller can connect to a display device, which can provide an image to a user based on activities performed by the system. In some embodiments, although not depicted, the system can include an edge device, which can be a cell phone and/or computer with a network enabled signal (e.g., cell signal), which can communicate information from the wellsite to a remote computing device for further analysis.

In some embodiments, the memory can include non-transitory computer-readable instructions that are executable by the processor to receive data from a wellsite sensor. In some embodiments, as discussed herein, the wellsite sensor can include a camera, which can include a standard image sensor and/or a thermal sensor. In some embodiments, the data obtained from the wellsite sensor can include image data that is captured with one or more of the cameras.

In some embodiments, the data can include image data from one or more images of the storage tank. As depicted in, phantom linerepresents a transition (e.g., interface) between oil stored in the tank and a gas (e.g., air). Oil is located below the lineand the gas is located above the line. In some embodiments, the image data obtained from the cameracan be used to determine a level of fluid in the storage tank, as discussed in U.S. application Ser. No./,, which is hereby incorporated by reference as though fully set forth herein.

In some embodiments, the camera-can be a thermal camera that is positioned such that the oil storage tankis within a field of view of the thermal camera. In some embodiments, the thermal camera can be connected, to the computer, which can analyze a feed from the thermal camera and determine a thermal gradation of the oil storage tankthat indicates a level of oil in the oil storage tank. For example, the portion of the oil storage tankfilled with oil can have a different temperature profile than an empty portion of the oil storage tankdue to the oil that fills the filled portion of the oil storage tank having a different temperature and/or heat capacity than air filling the empty portion of the oil storage tank.

In an example where the oil storage tankis half full, the bottom half containing oil may appear to have a different temperature (e.g., be colder) than the top half containing air, which can be represented in the feed received from the thermal camera by the computer. Through processing of the feed received from the thermal camera via computer vision, as discussed herein, a determination can be made by the computerand/or a central computer that the oil storage tank is half full. Likewise, a determination can be made of how full the oil storage tankis, at any level (e.g., 42% full, 59% full, 77% full, etc.). In some embodiments, computer vision can include algorithmic methods for processing image data for the purpose of extracting salient/contextually-relevant insights from that data. Generally computer vision can require no “training data” for the specific problem domain (e.g., increasing contrast on an image then analyzing pixels to find edges).

In an example, the camera-can be a thermal video sensor, which can be in communication with the computervia a wired or wireless connection. The thermal camera-can capture a thermal image of a storage tank. As can be seen in, the storage tankcan be partially filled with one or more liquids and/or gas. In some embodiments, the liquid(s) stored in the storage tankcan have a higher heat capacity than an air, which fills an empty portion of the tank. Accordingly, as a temperature in an environment in which the storage tankis located varies, the portion of the storage tankfilled with the liquid can be more resistant to a temperature change, due to its higher heat capacity. Accordingly, in an example where the environment in which the storage tankis located fluctuates in temperature, a temperature of an empty portion of the storage tank(e.g., the portion of the tank located above line) can be different than a filled portion of the storage tank (e.g., the portion of the tank located below line).

Additionally, the portion of the tank filled with liquids can contain liquids with differing heat capacities, which can produce different thermal profiles detectable by the camera-. For example, where the portion of the tank located between linesandcontains oil and the portion of the tank located below linecontains water, the two portions (e.g., oil containing portion between linesandand water containing portion below line) can have different thermal profiles due to the varying heat capacities of the two different liquids.

For instance, as a temperature increases throughout the day, a temperature of an empty portion of the tank can increase at a faster rate than a temperature of the filled portion, because a heat capacity of the liquid filling the filled portion is greater than a heat capacity of the air filling the empty portion. Accordingly, a thermal interfacecan form between the filled portion and the empty portion. In an example, the thermal interfacecan be a temperature differential between the filled portion and the empty portion.

In some embodiments, the thermal camera-can capture the thermal image of the storage tank. In an example, the data associated with the thermal image of the storage tankcan be provided to the computer, as discussed herein.

The computercan analyze the data and can determine a profile of the storage tank included in the thermal image of the storage tank. For example, an outline of the storage tankcan be determined in relation to a surrounding environment from the storage tank, which can be of a different temperature than the storage tankincluded in the thermal image of the storage tank. The profile of the storage tankcan be defined as an outline of the storage tank, which can be used in calculations made by the computerthat relate to how full the storage tankis.

In some embodiments, artificial intelligence can be used to make determinations associated with characteristics of the tank. In some embodiments, artificial intelligence and/or machine learning can include specifying a mathematical model to characterize a process of interest and using algorithmic techniques to fit training data to that model for the purposes of extracting salient/contextually-relevant insights in a domain of interest (e.g., training an object detection model to recognize a tank and associated thermal interfaces,,). For example, in some embodiments, data received from the camera-can be analyzed using computer vision to determine a characteristic associated with the wellsite. In some embodiments, the data received from the camera-can be analyzed using computer vision to determine a characteristic associated with the tank.

In an example, computer vision can be used to determine where the thermal interface,is located between the filled portion of the storage tankand the empty portion of the storage tank. In an example, the thermal interfaces,can be caused as a result of a changing temperature in an environment in which the tankis located. In an example, as the tank heats up or cools off throughout changing temperatures during the day or night, the thermal interfaces,can be caused due to the differing heat capacities of the fluids (e.g., air and liquid in the tank). For instance, a temperature of a liquid in the filled portion of the storage tankcan change at a slower rate than air in the empty portion of the storage tank.

Based on where the thermal interfaces,are located, the level of fluid in the storage tankcan be determined. In an example, a level of fluid in the storage tankcan be determined, based on a location of the thermal interfaces,between the filled portion of the storage tankand the empty portion of the storage tank. For instance, as depicted in, the thermal interfaceoccurs at a level that is roughly 0.1 of an overall height of the storage tankand the thermal interfaceoccurs at a level that is roughly 0.7 of an overall height of the storage tank. Thus, a determination can be made that liquid filling the filled portion of the tankoccupies roughly 0.7 of the total tank volume (e.g., 70 percent of the total tank volume). Furthermore, a determination can be made that 60 percent of the total tank volume is filled with oil and 10 percent of the total tank volume filled with water.

In some embodiments, a determination of the level of fluid in the storage tankcan be based on an area occupied by the liquid in a profile thermal image of the tank. For example, from a side profile view (e.g., two-dimensional) thermal image of the storage tank, an area of which is occupied by the liquid (e.g., filled portion of the storage tank) and/or air (e.g., empty portion of the storage tank) can be determined. A proportion of the area of which is occupied by the liquid and area of which is occupied by the air can be determined. For example, as depicted, in, roughly more than twice the volume of the storage tankis occupied by the liquid than that occupied by the air. Accordingly, a determination can be made that the storage tank is 70 percent full.

In some embodiments, the level of the tank can be computed by the computerand can be conveyed to a cloud computer or central computer. In some embodiments, data associated with the image taken by the camera-can be transmitted to the computer, a cloud computer, or central computer located offsite for one or more determinations of characteristics associated with the wellsite to be determined. As further discussed herein, in some embodiments, the image can include additional data. For example, the image can be labeled with additional data in some embodiments. In some embodiments, the image can be labeled with data that includes calculated levels associated with each of the thermal interfaces. For example, the image can be labeled with data that indicates that the thermal interfaceis located at 0.1 of a height of the storage tankand the thermal interfaceis located at 0.7 of a height of the storage tank. In some embodiments, the labeled image can be transmitted, as further discussed herein. The labeled image can be used for training purposes with respect to the use of artificial intelligence, as further discussed herein.

In some embodiments, as discussed herein, computer vision can be used to identify one or more characteristics associated with the storage tankthrough the associated image data thereof. In some embodiments, computer vision can include one or more algorithms that can be initially trained and continually updated to recognize the one or more characteristics associated with the storage tank. In some embodiments, computer vision can detect a change in pixel color between the interfaces,. For example, because of the differing heat capacities of the fluids stored in the storage tank, the interfaces,can generally be represented as a change in color, represented by the thermal image data.

In some embodiments, computer vision can be used to determine the location of the interfaces,, even when there is not a clear demarcation of where the interface is. In an example, in some embodiments, the interfaces,may not be defined lines that extend across the thermal image data. For instance, in some embodiments, the interfaces can be depicted on the thermal image as slow transitions from one color to another, creating a type of fuzzy transition and not a defined interface. In some embodiments, computer vision can be used and trained to identify transitions between the fluids, which are less than defined (e.g., are gradual transitions of color). In some embodiments, labeled images, as discussed herein, can be used for training to identify transitions between the fluids.

In some embodiments, the slow transition of color on the thermal image at the interfaces,can be caused by the dissipation of heat in the wall of the storage tank. For example, the fluids can transfer heat through a wall of the storage tank. As the heat is transferred through the wall of the storage tank, the heat can be dispersed in the material of the tank, which in some embodiments can be metal and/or plastic. In some embodiments, data can be initially trained by a human labeling of images and can be further refined with automated training of the model as more data and images are processed.

In some embodiments, the interface, which can be the interface between oil and water may not be a clear line due to an emulsion that exists between the two fluids (e.g., water and oil). In an example, computer vision can be trained to detect instances where emulsions exist and interpret the data associated with the thermal image to determine the data in a way that accounts for the emulsion in relation to the two fluids forming the emulsion. In an example, as depicted in, an emulsion exists between the interfaces,. In a physical sense, the interface between the oil and water can appear as a thicker line on the thermal image taken by the camera-, since the emulsion can have a heat capacity that is different than both oil and water. As depicted, the emulsion can exist between the interface lines,. In some embodiments, computer vision can be used to recognize that the emulsion exists and where the emulsion is located in order to accurately identify where the transitions between water, emulsion, and oil are located.

In some embodiments, computer vision can be used to detect extraneous data associated with the storage tank. For example, in some embodiments, changes in temperature along the surface of the tank can be caused by external factors, not associated with the differing heat capacities of fluids stored in the storage tank. For instance, in some embodiments, the sun can shine on select portions of the storage tank, causing some portions of the storage tank to be heated and some portions of the storage tankto not be heated.

In instances where the sun is rising or setting, other equipment on the well site and/or geographical features can cause portions of the storage tankto be shaded, while other portions of the storage tankare exposed to the sun. Accordingly, the portions of the storage tanklocated in the sun can be heated and the portions of the storage tankthat are located in the shade may not be heated. Thus, the thermal image of the tankcan depict a transition between the heated portion of the storage tankand the non-heated portion of the storage tank. In some embodiments, computer vision can determine the transition between the heated and non-heated portion of the storage tankand can recognize that the transition is a transition due to a temperature change caused by the sun and not a transition between fluids stored in the tank. Accordingly, embodiments of the present disclosure can classify and deal with extraneous data related to environmental conditions.

In an example, artificial intelligence can interpret the data obtained from the camera-and can make a determination, based on the data. In some embodiments, additional information can be provided to the system. In some embodiments, the additional information can include location information associated with a position of the wellsite geographically. In some embodiments, the additional information can include data associated with surrounding geography, vegetation, structures, etc., which may cast a shadow on the storage tank. In some embodiments, the additional information can include data associated with sun charts with respect to the times at which the sun sets and rises. In some embodiments, the additional information can include weather information associated with an area in which the wellsite is located. For example, the weather information can include wind speed, wind direction, temperature, humidity, amounts of precipitation, etc. In some embodiments, the additional information can include information with respect to a location of equipment on the wellsite, which can, for example cause shadows to be cast on the storage tank. In some embodiments, the additional information can include a location of the camera and direction in which the camera is facing, as well as a location and direction of the storage tankwith respect to the camera-.

In some embodiments, the additional information can be used by artificial intelligence to determine various factors associated with the storage tank, based on the additional information, which may be interpreted as one or more false interfaces between fluids in the storage tank. For example, in some embodiments, where a shadow is cast on the storage tank, artificial intelligence can be used to consider some or all of the additional information alongside the data obtained from the camera-to determine whether the interface is actually associated with the transition between fluids and not from a shadow being cast on the tank.

In some embodiments, computer vision can learn from previous patterns of data and utilize previous patterns with respect to making determinizations of characteristics on the wellsite. For example, if a shadow is cast on the storage tankevery day at a particular time, computer vision can be used to make a determination that data obtained as a result of the shadow being cast is not representative of an actual level of fluid in the storage tank.

For example, in some embodiments, the additional data can include a time of day and a time the sun rises, as well as the location of equipment that casts a shadow on the storage tank. As mentioned, in some embodiments, artificial intelligence can consider this data and can determine that a thermal transition is due to a shadow being cast on the storage tank and not an interface existing between fluids.

In some embodiments, artificial intelligence can determine how many interfaces are present and can determine that a detected thermal transition line is not an interface between fluids, based on the total number of thermal transition lines. For example, if four thermal transition lines are present, when only three thermal transition lines should be present, a determination can be made that one of the thermal transition lines is a false interface between fluids. In some embodiments, the computer vision can factor out the false data due to shadows on the tank, based on a rate of movement between the false shadow data and the other transition lines. For example, where the false shadow data represents a line moving in an opposite direction of the other transition lines present on the tank and/or represents a line moving at a much more rapid pace, due to a movement of the shadow, computer vision can disqualify that data from further analysis.

In some embodiments, artificial intelligence can be used to determine that one of the thermal transition lines is a false liquid interface, based on a geometry of the thermal transition line. For example, if the thermal transition line detected by computer vision is at an angle that is not a horizontal angle, a determination can be made that the thermal transition line is not due to a liquid interface. In some embodiments, computer vision can look for patterns in data associated with the liquid interfaces and use those patterns to disqualify extraneous image data that could be categorized as liquid interfaces by an untrained system. For example, if the camera-is located below a vertical level of a liquid interface, the thermal image can have a slight upward (e.g., convex) curve where the liquid interface is located. If the camera-is located at a same vertical level of the liquid interface, the thermal image can a straight line where the liquid interface is located. If the camera-is located above a vertical level of the liquid interface, the thermal image can have a slight downward (e.g., concave) curve.

In some embodiments, artificial intelligence can be used to generate various notification associated with the data collected form the camera-. In some embodiments, artificial intelligence can be used to determine a notification that a level of fluid in the storage tankis nearing a capacity of the storage tank. In some embodiments, artificial intelligence can be used to determine a notification that includes an indication of an amount of oil versus an amount of water in the storage tank. In some embodiments, an alert can be generated when a ratio of water to oil exceeds a particular ratio. In some embodiments, the alert can be indicative of a problem with the separator, thus indicating that the separatoris not effectively removing water from the oil, water mix passing through the separator from a wellhead.

As further depicted in, a second camera-can be configured to provide image data associated with a flare. Although the camerasare generally discussed herein as being directed towards a single piece of equipment and capturing image data of that single piece of equipment, camerasof the present disclosure can be directed towards one or more pieces of equipment, effectively allowing for them to capture image data associated with the one or more pieces of equipment.

In some embodiments, computer vision can be used to recognize particular features associate with the image data of the flare. In some embodiments, the camera-can include an image sensor and/or a thermal sensor that is configured to capture image data of a plume(e.g., smoke plume) generated by the flareand/or a flame (not depicted) generated by the flare. In some embodiments, computer vision can be used to determine a color and/or size of the plumeand/or flame generated by the flare. In some embodiments, the color of the plumegenerated by the flarecan be indicative of a health of the flare. For example, a plume that is black can indicate that the flare is running rich. In some embodiments, computer vision can be used to determine characteristics associated with the flare, even when extraneous data is included in the image captured with the camera-. For example, in some embodiments, a background of the image can include clouds, which can make it difficult to distinguish between a plumeassociated with the flare and the clouds. In some embodiments, the computer vision can be trained to recognize differences between the plumeand clouds based on motion associated with the plume, versus clouds located in a background of the image. In an example, the plumecan move with a faster apparent velocity than background clouds, since the plumecan be located closer to the camera-than the clouds.

In some embodiments, the camera-can include image data sensing functionality as well as thermal sensing functionality. Alternatively, the camera-can include more than one camera (e.g., two cameras), one of which includes an image data sensor and one of which includes a thermal sensor. In some embodiments, where clouds exist in a background of an image obtained with the camera-, a combination of a thermal image and an image data image can be used to determine characteristics associated with the flare. For example, where a color of the plumematches a color of the clouds, a combination of thermal mapping and image mapping with the two different types of cameras can be used to determine a size of the plume. The thermal data can first be used to locate a position and size of the plumeand the image data can be used for determining a health of the plume.

In some embodiments, artificial intelligence can be used to determine whether to generate a notification associated with the flarehealth, as further discussed herein. In some embodiments, artificial intelligence can be used to make determinations associated with characteristics of the flare(e.g., flare health). In some embodiments, artificial intelligence and/or machine learning can include specifying a mathematical model to characterize a process of interest and using algorithmic techniques to fit training data to that model for the purposes of extracting salient/contextually-relevant insights in a domain of interest (e.g., training an object detection model to recognize flares and black smoke from live camera streams given labeled training samples). For example, in some embodiments, data received from the camera-can be analyzed using computer vision to determine a characteristic associated with the wellsite. In some embodiments, the data received from the camera-can be analyzed using computer vision to determine a characteristic associated with the flare.

In some embodiments, a third camera-can be positioned, such that an image of one or more portions of piping are captured. For example, as depicted in, a portion of piping adjacent to the well headis within view of the camera-, as well as an associated leak (e.g., methane leak). In some embodiments of the present disclosure, computer vision can be used to analyze the image to determine that a methane leak exists. In some embodiments, the third camera can include an image sensor and/or thermal sensor. In some embodiments, the image sensor can be configured to detect a wavelength of light associated with the methane leak. Methane is absorbed throughout the infrared spectral range but has four fundamental zones centered around 2.3 μm, 3.3μm, 6.5 μm and 7.7 μm, with 3.3 μm and 7.7-7.8 μm showing the higher intensity. Although embodiments of the present disclosure are discussed in relation to a wellsite, embodiments of the present disclosure can also detect methane at other types of sites. For example, embodiments of the present disclosure can detect methane at a transfer station for natural gas, a waste management site, and/or a landfill. Further embodiments of the present disclosure can detect other types of fluids (e.g., gases and/or liquids) associated with other types of sites, such as ammonium nitrate vapor at a mine.

In further relation to, in some embodiments, the camera can be positioned such that a portion of the piping-of the wellsite is within view of the camera-. In some embodiments, the camera-can be positioned such that a substantial portion of the piping-,-, . . . ,-associated with the wellsite can be within view of the camera. Accordingly, the camera-can detect a leak among a majority of the piping-,-, . . . ,-associated with the wellsite. In some embodiments, a plurality of cameras can be disposed on the wellsite, such that each camera monitors a portion of the piping-,-, . . . ,-associated with the wellsite and together all or a substantial portion of the piping-,-, . . . ,-of the wellsite can be monitored.

In some embodiments, computer vision can be used to analyze a signal obtained from the camera-. In an example, computer vision can be used to look for characteristics of the image that fit characteristics associated with a fluid leak. In some embodiments, computer vision can detect leaks based on a difference in pixel color and/or intensity. For example, computer vision can be programmed with a baseline that does not include any leaks along the piping associated with the wellsite. When a leak is present, in some embodiments, computer vision can be used to detect the leak, based on a change in pixels of the image captured by the camera-. In some embodiments, a fluid leak can be associated with a cluster of pixels that differ in color and/or intensity versus those pixels in a baseline image.

In some embodiments, computer vision can be used to detect false positives associated with an image captured with the camera-. For example, computer vision can be used to detect the difference between a fluid leak and other types of anomalies, such as pooled water on the ground due to rain, etc.

In some embodiments, artificial intelligence can be used to determine what type of notification to generate. In some embodiments, artificial intelligence can be used to make a decision of whether to generate a notification and what type of notification to generate. In some embodiments, artificial intelligence can be used to make a determination of a size of the leak, based on a number of pixels associated with the leak.

Although some embodiments of the present disclosure are discussed for use in relation to a wellsite, some embodiments can be used in relation to other types of sites. For example, some embodiments of the present disclosure can be used in relation to a transfer station for natural gas, a waste management site, and/or a landfill for the detection of methane gas. In some embodiments, embodiments of the present disclosure can be used in relation to a mining site. For example, embodiments of the present disclosure can be used to monitor for vapors released from ammonium nitrate. In an example, embodiments of the present disclosure can be used to monitor for vapors released from ammonium nitrate stored in a silo at a mine. For example, in some embodiments, one or more cameras (e.g., optical gas imagers) can be disposed in proximity to the silo and/or around the mine, such that a field of view of the cameras can capture any areas where there may be a likelihood of supposed vapors released from ammonium nitrate. In some embodiments, the camera can include an image sensor and/or thermal sensor. In some embodiments, the image sensor can be configured to detect a wavelength of light associated with the ammonium nitrate leak. The image sensor can be configured to detect a wavelength centered around 3.3 μm and 7.4 μm.