<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Jeff Kimmel on Medium]]></title>
        <description><![CDATA[Stories by Jeff Kimmel on Medium]]></description>
        <link>https://medium.com/@kimmelj?source=rss-13d5c3943eef------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/0*Y4sLUsG-XJm8Xfww.jpg</url>
            <title>Stories by Jeff Kimmel on Medium</title>
            <link>https://medium.com/@kimmelj?source=rss-13d5c3943eef------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Mon, 16 Nov 2026 21:12:47 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@kimmelj/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[Avoiding the Perceived Obstacles of Fault Detection and Diagnostics]]></title>
            <link>https://medium.com/elipsa/avoiding-the-perceived-obstacles-of-fault-detection-and-diagnostics-5d3d5ae15006?source=rss-13d5c3943eef------2</link>
            <guid isPermaLink="false">https://medium.com/p/5d3d5ae15006</guid>
            <category><![CDATA[smart-buildings]]></category>
            <category><![CDATA[hvac]]></category>
            <category><![CDATA[fault-detection]]></category>
            <dc:creator><![CDATA[Jeff Kimmel]]></dc:creator>
            <pubDate>Mon, 24 Jun 2024 18:09:56 GMT</pubDate>
            <atom:updated>2024-06-24T18:09:56.420Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/740/0*pDCnioQsGLexsoWu.jpg" /></figure><p>Implementing Fault Detection and Diagnostics (FDD) in your buildings can help reduce energy consumption and cut utility bills. In addition, it can improve indoor air quality and comfort, increase equipment lifespan, and operational efficiencies. However, despite the benefits, many are reluctant to move forward with implementation. Why? What are the perceived obstacles in implementing FDD in buildings?</p><p>Perceived obstacles include high initial costs, complexity and technical challenge, data management and quality, skill and expertise requirements, resistance to change, and ROI uncertainty.</p><p>Let’s tackle these and explain how Elipsa’s innovative approach helps remove these obstacles and enable implementation at scale.</p><h3>High Initial Costs</h3><p>Elipsa’s FDD solution is built off of it’s automated machine learning platform. Through the use of AI, Elipsa automates a significant portion of the implementation process. This automation means less setup time and therefore lower setup fees.</p><h3>Auto-Tagging</h3><p>In order to implement FDD you need to have equipment and points. For most systems, it takes a considerable amount of time to gather this information. Then it takes even more time to configure the equipment and associated points in an FDD system even before configuring what you will actually be monitoring.</p><p>Elipsa’s auto-tagging fixes this. With auto-tagging, all you need to do is export a list of your equipment and points. Using AI, Elipsa can then automatically determine the type of equipment and points configuring the FDD system in the process. This automated approach drastically reduces the time and cost to configure the FDD system with the equipment that will eventually be monitored.</p><h3>Skill and Expertise Requirements</h3><p>With equipment and points now configured in the FDD system, what are you going to monitor? For some FDD systems, you need to answer that question yourself. Some systems require you to code individual rules. Others have allow you to pick from a list. Both of these approaches requires a lot of time and knowledge. Elipsa takes a different approach.</p><h3>AI Monitoring Templates</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/740/0*knAZZtnsjRK0Kep_.png" /></figure><p>Imagine a menu of all possible components to monitor for a given piece of equipment. Each of those have a set of points to be monitored. So, you know the points you need and with the system already configured you know what points you have. With Elipsa, don’t bother selecting what you want to monitor, the system chooses for you. Automatically apply monitoring for all components that you have a sufficient number of points for. Gone is the need for expertise of deciding which points to monitor and which components you have or do not.</p><p>Pure rules based systems monitor equipment based on defined thresholds. You can utilize industry standard thresholds or select your own. This is a daunting task as you think about the level of experience needed to appropriately come up with the thresholds to monitor for.</p><p>With an AI-based approach there is no need for thresholds. The AI learns the thresholds. More specifically, the AI learns normal operation to detect when the system is trending away from normal. The end result is once again a much faster implementation but also removed is the obstacle of having highly skilled and experienced personnel doing the implementation.</p><h3>Data Management and Quality</h3><p>For advanced analytics, data is always a challenge and particularly access to that data. Elipsa utilizes MQTT to integrate into numerous systems. In addition, we have partnered with companies such as Tridium to directly access data from your Building Management System.</p><p>There are certainly buildings that are not connected. Installing sensors could certainly be cost prohibitive but it also poses the question of when do you have enough data. AI needs historical data to build a model. So, what do you do if you do not have a long history of quality data?</p><p>Elipsa’s templates can get you started without historical data. In fact, if builds your historical data set as the equipment runs. Elipsa’s instantly employs ASHRAE 36 rules to be able to monitor equipment within an hour of streaming equipment data to Elipsa. Within a few short weeks, enough data will have been received to build an AI monitoring model for your equipment’s components. This model creation is done automatically without additional configuration by the user.</p><p>Connectivity and quality are certainly issues but Elipsa has taken a partnership approach to go directly to the source and remove that burden from you.</p><h3>Uncertain ROI</h3><p>According to the US Department of Energy, buildings can save 3–9% per year on energy costs by implementing FDD. The challenge is that where you are on the range depends a lot on your building. From age of building and equipment to energy conservation measures that have been implemented. The savings range is fairly easy to calculate but the second piece of your return on investment is the investment. This ties to the obstacles if initial costs and skilled employees to implement. Those are often hard to calculate and as we said earlier, often considered too high.</p><p>The large amount of money and time to implement makes it hard to see a justifiable ROI. However, if we can eliminate the complexities of implementation and in turn greatly lower the cost, that ROI becomes much more attainable.</p><h3>Resistance to Change</h3><p>This is an industry where things have been run the same way for a long time. One can easily say that the resistance to implementing FDD is due to resistance to change. Really though, that is a catch all. With all the other perceived obstacles it’s easier to just avoid them and say you don’t want change. In general though, change is easier to accept when it is easy. By eliminating many of the obstacles discussed earlier, FDD is now easier to try and therefore it is easier to change.</p><p>We are at a point where change is now possible and hopefully soon to a point where it is more acceptable with Elipsa’s FDD solution.</p><h3>Conclusion</h3><p>There are obstacles in implementing any technology, especially new technology. However, when something comes along that helps you avoid those obstacles you can finally implement at scale and reap the rewards. Elipsa’s platform is built on AI and built on simplicity. This simplicity finally makes FDD, and the savings associated with it, a possibility paving the way for greater operational efficiencies and energy saving.</p><p>For more information contact us at <a href="/service/mailto:info@elipsa.ai">info@elipsa.ai</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=5d3d5ae15006" width="1" height="1" alt=""><hr><p><a href="https://medium.com/elipsa/avoiding-the-perceived-obstacles-of-fault-detection-and-diagnostics-5d3d5ae15006">Avoiding the Perceived Obstacles of Fault Detection and Diagnostics</a> was originally published in <a href="https://medium.com/elipsa">Elipsa</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[ASHRAE 36 Fault Detection Made Simple]]></title>
            <link>https://medium.com/elipsa/ashrae-36-fault-detection-made-simple-156f8623fa98?source=rss-13d5c3943eef------2</link>
            <guid isPermaLink="false">https://medium.com/p/156f8623fa98</guid>
            <category><![CDATA[fault-detection]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[ashrae]]></category>
            <category><![CDATA[smart-buildings]]></category>
            <dc:creator><![CDATA[Jeff Kimmel]]></dc:creator>
            <pubDate>Tue, 18 Jun 2024 15:29:52 GMT</pubDate>
            <atom:updated>2024-06-18T15:29:52.300Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/714/1*T-3j5jKdQ5bvo6gSZbow-Q.png" /></figure><p>Pretty much everyone we talk to in the building space knows ASHRAE 36, the guidelines for <strong>High-Performance of Operations for HVAC Systems</strong>.</p><p>If you open the document and look at the scope of the document it lists:</p><ol><li>This guideline provides detailed sequences of operation for HVAC systems</li><li>This guideline describes functional tests that, when performed, will confirm implementation of the sequences of operation</li></ol><p>So, it is no surprise that when you ask people what ASHRAE 36 encompasses, the answer is control sequences to optimize the performance of your HVAC systems. What many do not realize is within the same guideline is a list of rules for continuously monitoring the performance of these HVAC systems known as Automatic Fault Detection and Diagnostics (AFDD).</p><h3>ASHRAE 36 AFDD</h3><p>These rules offer insights into which elements are most important to track and monitor in order to ensure that your systems are running optimally. They offer a best of breed approach to detecting potential faults.</p><blockquote><em>many think the implementation of these rules are unattainable</em></blockquote><p>However, for those that we talk to who do understand that ASHRAE 36 goes beyond control sequences and offers AFDD, many think the implementation of these rules are unattainable.</p><p>Put another way, they often comment that the rules cannot be implemented if you also do no take the time to implement the control sequences themselves.</p><p>A rules based approach typically monitors against a threshold. For example, when monitoring an AHU, the guidelines say to alert someone when the current duct static pressure is more than 25 Pa (or 0.1&quot;) below the average duct static pressure. 25 Pa was deemed to be the appropriate threshold after extensive testing but is 25 Pa appropriate for all AHUs and is it attainable if you do not implement the defined AHU control sequences on that device?</p><h3>AI-Based Monitoring Built Off of ASHRAE 36</h3><p>This is the challenge people see with ASHRAE 36 and the overall challenge with rules based approaches to FDD. So how do we overcome this?</p><p>This is where AI comes in. At Elipsa, we follow the same guidelines of ASHRAE 36 in terms of points being monitored for specific equipment types and components. However, with the use of AI, we do not need to set a rigid threshold and apply it across all equipment. Instead, the AI learns what normal operation of a given piece of equipment is in order to then monitor for abnormalities. In other words, it is similar to learning what the appropriate threshold is for each and every piece of equipment. So, an AHU in Austin could in theory have a different alerting point than an AHU in Boston.</p><p>Based on this approach, ASHRAE 36 guidelines for AFDD can be implemented across the board with no concerns of whether your thresholds are appropriate or attainable.</p><p>Implementing AFDD has been proven to reduce maintenance costs and reduce energy consumption by ensuring higher levels of system performance. AI and Elipsa help make that more attainable and easier than ever before to start optimizing the operation of your critical equipment.</p><p>To learn more, contact us at <a href="/service/mailto:info@eipsa.ai">info@eipsa.ai</a> or visit <a href="https://www.elipsa.ai/">https://www.elipsa.ai</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=156f8623fa98" width="1" height="1" alt=""><hr><p><a href="https://medium.com/elipsa/ashrae-36-fault-detection-made-simple-156f8623fa98">ASHRAE 36 Fault Detection Made Simple</a> was originally published in <a href="https://medium.com/elipsa">Elipsa</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The How-To on Haystack Auto-Tagging with Elipsa]]></title>
            <link>https://medium.com/elipsa/the-how-to-of-haystack-auto-tagging-with-elipsa-b2d1dbed1211?source=rss-13d5c3943eef------2</link>
            <guid isPermaLink="false">https://medium.com/p/b2d1dbed1211</guid>
            <category><![CDATA[haystack]]></category>
            <category><![CDATA[smart-buildings]]></category>
            <category><![CDATA[hvac]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[ai]]></category>
            <dc:creator><![CDATA[Jeff Kimmel]]></dc:creator>
            <pubDate>Mon, 10 Jun 2024 15:38:52 GMT</pubDate>
            <atom:updated>2024-06-10T15:44:29.283Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/740/0*2UkPSgPPyOaOAAa1.jpg" /><figcaption>The no-code approach to tagging building data</figcaption></figure><p>With Elipsa’s auto-tagging capabilities, you can go from a list of equipment and point names to fully tagged equipment and points in a matter of minutes. Starting with a CSV file, you can import your information to be processed by Elipsa’s AI platform.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/740/0*ly1tqDPAYWuzhH4J.png" /></figure><p>The sensitivity of the AI can be adjusted, with zero needing a near perfect match to an equipment or point name that the system has seen before and 10 providing a prediction even if it is a fairly low confidence.</p><p>Similar to how a human would look at RTU1 and know it is an RTU or VAV_102 and know it is a VAV, AI can make similar predictions. If the AI is wrong, you can utilize dropdowns to choose the correct type. Once equipment and points are added, the appropriate haystack tags are automatically applied. Additionally, the AI learning is updated to reinforce what it got right and correct what it got wrong.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/740/0*LnU7cEjNX9q6IXrX.png" /></figure><p>For predicting point, we divide the equipment by type to make it easier to manage.</p><p>Looking at the RTUs, we show the imported points in orange and the mapped points in the dropdown. For equipment such as RTUs with multiple compressors, an AHU with multiple supply fans, or a chiller plant where all points are tied to chiller plant in your BMS rather than the individual equipment, you can add sub-equipment easily in the interface to create a connected system.</p><p>Similar to adding equipment, if the AI makes an incorrect prediction, you can choose the correct point in the dropdown. Saving the individual equipment will add the point to the learning. Refreshing the predictions will then be able to make an accurate prediction for all other equipment with similar named points.</p><p>Once we save an equipment type, the color indicator changes to green allowing us to keep track of what points have been added.</p><p>With equipment and points now mapped and saved, we can utilize the haystack tags to run features of the Elipsa platform just as fault detection and diagnostics or we can export the points to feed back into your BMS.</p><p><a href="https://youtu.be/U1DFu5FPQZY">https://youtu.be/U1DFu5FPQZY</a></p><p>To learn more, contact us at <a href="/service/mailto:info@elipsa.ai">info@elipsa.ai</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=b2d1dbed1211" width="1" height="1" alt=""><hr><p><a href="https://medium.com/elipsa/the-how-to-of-haystack-auto-tagging-with-elipsa-b2d1dbed1211">The How-To on Haystack Auto-Tagging with Elipsa</a> was originally published in <a href="https://medium.com/elipsa">Elipsa</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Tag Your Building In Record Time: Simple And Scalable AI]]></title>
            <link>https://medium.com/elipsa/tag-your-building-in-record-time-simple-and-scalable-ai-df8c1625e8f8?source=rss-13d5c3943eef------2</link>
            <guid isPermaLink="false">https://medium.com/p/df8c1625e8f8</guid>
            <category><![CDATA[smart-buildings]]></category>
            <category><![CDATA[hvac]]></category>
            <category><![CDATA[auto-tagging]]></category>
            <category><![CDATA[predictive-maintenance]]></category>
            <dc:creator><![CDATA[Jeff Kimmel]]></dc:creator>
            <pubDate>Mon, 03 Jun 2024 17:32:12 GMT</pubDate>
            <atom:updated>2024-06-03T17:32:12.461Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/740/0*AWHT2mG42TguHs3M.png" /><figcaption>source: acentech.com</figcaption></figure><p>Tagging building HVAC data, with a tagging ontology such as haystack, is critical for enhancing the functionality, efficiency, and effectiveness of HVAC systems within a building. By improving data organization, facilitating predictive maintenance, enabling better integration with building management systems, and supporting advanced analytics, tagging contributes to significant operational improvements, energy savings, and enhanced occupant comfort.</p><h3>Let’s paint the picture of a perfect building:</h3><p>1) You have vast amounts of data that is well organized making it a cinch to retrieve and manage specific information. You instantly know the type of every piece of equipment and each of their data points such as temperatures and pressures</p><p>2) You have detailed real-time monitoring and diagnostics on your systems. Quickly identify issues such as abnormal temperatures or faulty components</p><p>3) Your equipment is operating efficiently ensuring low costs and energy consumption. HVAC operation is fully optimized identifying patterns of inefficiencies and automatically adjusting controls to reduce operational costs</p><p>4) Advanced analytics such as machine learning is proactively monitoring for issues before they occur. Predict component lifespan and prevent unexpected downtime.</p><p>5) Enhance your equipment’s integration with your building management system. Seamlessly connect all equipment and data to your BMS to enhance your overall control and automation. All of your data is standardized and normalized across your equipment and buildings</p><p>6) Full reporting and compliance. Your detailed reports ensure compliance with regulatory requirements and industry standards. Seamlessly retrieve data for audits and regulatory inquiries.</p><p>7) You provide the highest level of user experience and comfort. Automatically adjust settings based on current conditions and occupancy to provide optimal comfort levels across various areas of your building</p><p>8) Your building is generating advanced insights. Apply advanced analytics to gain a deeper understanding of your equipment. Identify trends and anomalies showing areas of improvement and insights for building managers</p><h3>Tagging Unlocks Your Building’s Potential. So Why Don’t People Do It?</h3><p>All of this is possible today and easy to implement if you do one thing, tag your data.</p><p>With a data model such as project haystack, you can add standardized meta data to your equipment and points to unlock the tremendous amount of value stored within your building.</p><p>All of this value that we just laid out can be yours. But more often than not the response we hear is “nobody tags”.</p><p>On one hand, this utopian building is at your fingertips thanks to tagging. So what is the road block? What is the reason behind the answer “nobody tags”? The answer is time.</p><p>The time it takes you or your system integrator to add these tags to your data is far too cost prohibitive that it’s nearly impossible to expect a return on that investment. As a building owner or a systems integrator, if you had a proposal in front of you that included tagging and one that would not, it’s nearly impossible to get over the sticker shock of the time and materials needed to tag for you to fully see the value at the end of it.</p><p>So, nobody tags.</p><p>But what if time was not an issue? What if you could do it for half the cost, a tenth of the cost or even less? What if you could do this because it was not your SI pouring over a screen for hours but instead it was AI analyzing your building in seconds?</p><h3>Auto-Tag Your Building Data with AI</h3><p>Artificial Intelligence is meant to mimic human intelligence. One such area is natural language processing and the ability to understand language and context.</p><p>A human could look at an equipment name in your BMS such as AHU_Lobby or Chlr1 and reasonably determine that you are looking at an AHU and a Chiller. Similarly, you can safely assume that DATemp in a VAV is discharge air temperature or FanStatus is, well Fan Status. AI can be utilized to do the same.</p><p>One big problem across the industry is that point names are hardly ever consistent. That is true across the same building, never mind across different buildings programmed by</p><p>different engineers. So, that is where tagging comes in to have a standard set of normalized meta data to help applications know what equipment and points are.</p><p>AI can cut through the grunt work. Using knowledge gained from previous buildings, AI can predict the equipment and point types similar to how a human would. Once the type is known we can automatically assign standard tags to the equipment and points. With the assistance of AI, the process of tagging thousands of points, which used to take hours, can now take minutes.</p><p>The time hurdle has been eliminated and the fully optimized and monitored building that we described earlier is now a possibility.</p><p>To learn more, and to see how auto tagging works, contact us at <a href="/service/mailto:info@elipsa.ai">info@elipsa.ai</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=df8c1625e8f8" width="1" height="1" alt=""><hr><p><a href="https://medium.com/elipsa/tag-your-building-in-record-time-simple-and-scalable-ai-df8c1625e8f8">Tag Your Building In Record Time: Simple And Scalable AI</a> was originally published in <a href="https://medium.com/elipsa">Elipsa</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Tag Your Building In Record Time: Simple And Scalable AI]]></title>
            <link>https://kimmelj.medium.com/tag-your-building-in-record-time-simple-and-scalable-ai-5d299f1332de?source=rss-13d5c3943eef------2</link>
            <guid isPermaLink="false">https://medium.com/p/5d299f1332de</guid>
            <category><![CDATA[predictive-maintenance]]></category>
            <category><![CDATA[smart-buildings]]></category>
            <category><![CDATA[auto-tagging]]></category>
            <category><![CDATA[hvac]]></category>
            <dc:creator><![CDATA[Jeff Kimmel]]></dc:creator>
            <pubDate>Mon, 03 Jun 2024 17:29:53 GMT</pubDate>
            <atom:updated>2024-06-03T17:29:53.260Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/740/0*m-FKroUv1LAYcuE8.png" /><figcaption>source: acentech.com</figcaption></figure><p>Tagging building HVAC data, with a tagging ontology such as haystack, is critical for enhancing the functionality, efficiency, and effectiveness of HVAC systems within a building. By improving data organization, facilitating predictive maintenance, enabling better integration with building management systems, and supporting advanced analytics, tagging contributes to significant operational improvements, energy savings, and enhanced occupant comfort.</p><h3>Let’s paint the picture of a perfect building:</h3><p>1) You have vast amounts of data that is well organized making it a cinch to retrieve and manage specific information. You instantly know the type of every piece of equipment and each of their data points such as temperatures and pressures</p><p>2) You have detailed real-time monitoring and diagnostics on your systems. Quickly identify issues such as abnormal temperatures or faulty components</p><p>3) Your equipment is operating efficiently ensuring low costs and energy consumption. HVAC operation is fully optimized identifying patterns of inefficiencies and automatically adjusting controls to reduce operational costs</p><p>4) Advanced analytics such as machine learning is proactively monitoring for issues before they occur. Predict component lifespan and prevent unexpected downtime.</p><p>5) Enhance your equipment’s integration with your building management system. Seamlessly connect all equipment and data to your BMS to enhance your overall control and automation. All of your data is standardized and normalized across your equipment and buildings</p><p>6) Full reporting and compliance. Your detailed reports ensure compliance with regulatory requirements and industry standards. Seamlessly retrieve data for audits and regulatory inquiries.</p><p>7) You provide the highest level of user experience and comfort. Automatically adjust settings based on current conditions and occupancy to provide optimal comfort levels across various areas of your building</p><p>8) Your building is generating advanced insights. Apply advanced analytics to gain a deeper understanding of your equipment. Identify trends and anomalies showing areas of improvement and insights for building managers</p><h3>Tagging Unlocks Your Building’s Potential. So Why Don’t People Do It?</h3><p>All of this is possible today and easy to implement if you do one thing, tag your data.</p><p>With a data model such as project haystack, you can add standardized meta data to your equipment and points to unlock the tremendous amount of value stored within your building.</p><p>All of this value that we just laid out can be yours. But more often than not the response we hear is “nobody tags”.</p><p>On one hand, this utopian building is at your fingertips thanks to tagging. So what is the road block? What is the reason behind the answer “nobody tags”? The answer is time.</p><p>The time it takes you or your system integrator to add these tags to your data is far too cost prohibitive that it’s nearly impossible to expect a return on that investment. As a building owner or a systems integrator, if you had a proposal in front of you that included tagging and one that would not, it’s nearly impossible to get over the sticker shock of the time and materials needed to tag for you to fully see the value at the end of it.</p><p>So, nobody tags.</p><p>But what if time was not an issue? What if you could do it for half the cost, a tenth of the cost or even less? What if you could do this because it was not your SI pouring over a screen for hours but instead it was AI analyzing your building in seconds?</p><h3>Auto-Tag Your Building Data with AI</h3><p>Artificial Intelligence is meant to mimic human intelligence. One such area is natural language processing and the ability to understand language and context.</p><p>A human could look at an equipment name in your BMS such as AHU_Lobby or Chlr1 and reasonably determine that you are looking at an AHU and a Chiller. Similarly, you can safely assume that DATemp in a VAV is discharge air temperature or FanStatus is, well Fan Status. AI can be utilized to do the same.</p><p>One big problem across the industry is that point names are hardly ever consistent. That is true across the same building, never mind across different buildings programmed by</p><p>different engineers. So, that is where tagging comes in to have a standard set of normalized meta data to help applications know what equipment and points are.</p><p>AI can cut through the grunt work. Using knowledge gained from previous buildings, AI can predict the equipment and point types similar to how a human would. Once the type is known we can automatically assign standard tags to the equipment and points. With the assistance of AI, the process of tagging thousands of points, which used to take hours, can now take minutes.</p><p>The time hurdle has been eliminated and the fully optimized and monitored building that we described earlier is now a possibility.</p><p>To learn more, and to see how auto tagging works, contact us at <a href="/service/mailto:info@elipsa.ai">info@elipsa.ai</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=5d299f1332de" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Optimize Oil Drilling Rates with AI]]></title>
            <link>https://medium.com/elipsa/optimize-oil-drilling-rates-with-ai-b27f9bccec52?source=rss-13d5c3943eef------2</link>
            <guid isPermaLink="false">https://medium.com/p/b27f9bccec52</guid>
            <category><![CDATA[upstream]]></category>
            <category><![CDATA[optimization]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[oil-and-gas]]></category>
            <category><![CDATA[ai]]></category>
            <dc:creator><![CDATA[Jeff Kimmel]]></dc:creator>
            <pubDate>Tue, 03 Jan 2023 14:27:09 GMT</pubDate>
            <atom:updated>2023-01-03T14:27:09.397Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*2xALiEkRFwTC6Pgq" /><figcaption>Photo by <a href="https://unsplash.com/@luandmario?utm_source=medium&amp;utm_medium=referral">Maria Lupan</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>Drilling is critical in the development of oil and gas resources. Drilling efficiently is a critical driver to reducing economic loss as well as reducing the environmental impact from such drilling.</p><p>One of the key components of being able to drill efficiently is understanding the rate of penetration or ROP. <strong>Onshore drilling rates range from $200,000 to $310,000 per day while offshore rates could reach up to $800,000 per day</strong>. Maximizing your ROP reduces the number of days required to drill which reduces the overall cost while also reducing the indirect impact to the land and environment as a whole.</p><h3>Data</h3><p>A large volume of drilling data is commonly collected by service companies during the drilling process. In addition, petrophysical logs are often recorded as well. The rate at which a drill can cut through a material such as rock is a function of the drill and material itself. For example, drilling through sandstone would result in a different expected ROP than drilling through limestone. Predicting the soil and rock consistency can be highly valuable to the drilling process. This will be addressed in a future post, but for purposes of this article, we will focus on accurately predicting the ROP based solely on drilling data.</p><p>When looking at the features of drilling data, the ROP of the drill is a direct function of drilling attributes such as the surface torque, the weight on the bit, weight on hook, as well as information about the drill mud and bore hole such as pump time, stand pipe pressure, mud flow in. While physics can help determine the data points or features that should directly impact ROP, machine learning is what allows us to analyze patterns within these data points.</p><h3>Machine Learning</h3><p>In this particular use case, we are seeking to predict a value or specifically the value of ROP. As such, in machine learning, this would utilize a regression algorithm.</p><p>Taking asingle run worth of drilling data from the Volve drilling data set, we extracted the key features above as well as the known ROP to learn from. We uploaded the data to Elipsa choosing the ROP column as the values to learn, and the remaining data points as the features or data to learn from.</p><p>Elipsa’s no-code platform runs through an automated data science experiment performing tasks such as data splitting, normalization, etc. and then builds a series of machine learning models on the well data. Testing data is then automatically run against the model and resulting models are sorted based on the predictive results on this unseen data.</p><p>In our example, a Decision Tree algorithm proved to be most accurate at predicting the ROP values. The results show how close predictions were inside specific thresholds as well as showing which data points are most important to driving the success of the final model.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/975/1*w5a-6xsfI34hJlWTGkXihA.png" /><figcaption>Elipsa Results</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/975/1*Ped_sXJTgqjyYIHHmGTjxg.png" /><figcaption>Elipsa Data Importance</figcaption></figure><p>Making predictions against this new model is as simple as an API call, allowing your existing data systems to become AI enabled.</p><p>Taking a new wellsite log, we run the data against the model to predict ROP. As you can see from the results, the majority of the predicted results are inline with actual.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/975/1*llW2tnLU4Ps39qNJagWhPQ.png" /></figure><h3>ROP Optimization</h3><p>As we have seen, through machine learning, we are able to accurately predict the drills ROP based on drilling and mud log data. However, those logs also include the current ROP. So, you may be asking where the value is in predicting something that is already known in real-time.</p><p>To extract the most value from these predictions, we are not looking solely to predict the ROP based on current data values but to understand how the ROP will change as values change. This focus then enables us to understand what changes can be made to maximize ROP.</p><p>Certain data points used by our model are a function of the environment surrounding the bit. While those give a great deal of information helpful in predicting the ROP, we cannot change them. However, if we can take the current data points along with data points that we can change, such as Torque, we can optimize the drill configuration to result in the highest ROP for the current conditions.</p><p>The result is now not simply a prediction but a controlled change to help drill faster based on real-time conditions.</p><h3>Summary</h3><p>Well drilling is one of the most expensive and environmentally taxing aspects of the upstream oil and gas industry. Machine learning can help companies to increate their drilling rates by taking in real-time information about mud logs and drilling in order to optimize the drilling configuration. The result can have a staggering impact on the economic success and environmental impact of an oil well.</p><p><em>Elipsa is a leader in delivering fully automated, scalable AI solutions for industrial IoT applications. Elipsa’s Intelligent Monitoring Platform seamlessly deploys across any workflow on the edge or in the cloud, increasing the availability and output of critical equipment. Elipsa’s self-training AI models and bolt-on approach enable AI deployments that are simple, fast, and flexible. To learn more, please visit </em><a href="https://www.elipsa.ai/"><em>https://www.elipsa.ai</em></a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=b27f9bccec52" width="1" height="1" alt=""><hr><p><a href="https://medium.com/elipsa/optimize-oil-drilling-rates-with-ai-b27f9bccec52">Optimize Oil Drilling Rates with AI</a> was originally published in <a href="https://medium.com/elipsa">Elipsa</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Can Machine Learning Reduce Flaring?]]></title>
            <link>https://medium.com/elipsa/can-machine-learning-reduce-flaring-ea59aacc4332?source=rss-13d5c3943eef------2</link>
            <guid isPermaLink="false">https://medium.com/p/ea59aacc4332</guid>
            <category><![CDATA[ai-for-good]]></category>
            <category><![CDATA[oil-and-gas]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[machine-learning]]></category>
            <dc:creator><![CDATA[Jeff Kimmel]]></dc:creator>
            <pubDate>Tue, 13 Dec 2022 18:11:02 GMT</pubDate>
            <atom:updated>2022-12-13T18:12:08.675Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*QwJW5NLkGBqTBN6e1eqEEA.png" /><figcaption>iea.org</figcaption></figure><p>How Predictive Maintenance can help reduce flaring across the Oil and Gas industry.</p><h3>What is flaring?</h3><p>In the production of Oil and Gas, flaring is the process of burning off excess or otherwise unwanted gases. Typically, methane needs to be disposed of and flaring is a technique of disposing of the gas via combustion.</p><p>Methane is known to have a 25X impact over CO2 on the environment. Flaring burns off the unwanted methane which produces CO2 and water into the atmosphere. However, while flaring converts methane to the lesser of two evils in the form of CO2, the flaring process does not fully burn off all methane levels and thus the process is still a key contributor to the release of both CO2 and methane which are the leading causes of climate change.</p><h3>What is Predictive Maintenance?</h3><p>Predictive maintenance techniques utilize machine data and machine learning to help determine the condition of in-service equipment in order to estimate when maintenance should be performed.</p><p>The benefits of predictive maintenance are typically centered around extending the useful life of critical equipment while also saving on cost of maintenance and parts due to efficiencies gained in the maintenance schedule.</p><p>However, there are often overlooked indirect benefits as well.</p><h3>Predictive Maintenance’s Impact on Flaring</h3><p>In the production of oil and gas, drilling equipment and the wells themselves must be monitored and maintained. Employing predictive maintenance can optimize the maintenance schedule to perform critical maintenance only when it is needed and not strictly based on a schedule. In addition, early indicators of failure allow for adjustments to be made to equipment to avoid failure and push off maintenance. The indirect benefit to this is how the optimized schedules and extended equipment useability impact flaring.</p><p>Upstream companies are tasked with cutting down on flaring to fight climate change but some amount of flaring is routine and necessary.</p><p>When maintenance is to be performed on a well, the well must first be depressurized for safety. This process involves burning off the excess gas via flaring and in turn releasing pollutants into the atmosphere. While there are technologies in use and more in development to make use of the methane in ways that don’t require flaring, the reality is that flaring is a necessary part of oil production around the world.</p><p>However, even though flaring is necessary, it doesn’t mean that we cannot reduce it.</p><p>As noted, predictive maintenance optimizes the maintenance schedule and extends the usability of machines to reduce downtime. This optimization reduces the number of times maintenance is to be performed. This in turn reduces the number of times that maintenance based flaring needs to occurs.</p><h3>Summary</h3><p>Through the use of machine learning, Predictive Maintenance benefits go beyond extending the life of equipment and has the potential to have a positive impact on the environment as a whole. By deploying AI-based solutions, upstream oil and gas companies can reap the cost saving benefits of predictive maintenance while reducing their carbon footprint in the process.</p><p><em>Elipsa is a leader in delivering fully automated, scalable AI solutions for industrial IoT applications. Elipsa’s Intelligent Monitoring Platform seamlessly deploys across any workflow on the edge or in the cloud, increasing the availability and output of critical equipment. Elipsa’s self-training AI models and bolt-on approach enable AI deployments that are simple, fast, and flexible. To learn more, please visit </em><a href="https://www.elipsa.ai/"><em>https://www.elipsa.ai</em></a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=ea59aacc4332" width="1" height="1" alt=""><hr><p><a href="https://medium.com/elipsa/can-machine-learning-reduce-flaring-ea59aacc4332">Can Machine Learning Reduce Flaring?</a> was originally published in <a href="https://medium.com/elipsa">Elipsa</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[AI Spotlight: Leak Detection]]></title>
            <link>https://medium.com/elipsa/ai-spotlight-leak-detection-7e50f346f5b1?source=rss-13d5c3943eef------2</link>
            <guid isPermaLink="false">https://medium.com/p/7e50f346f5b1</guid>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[oil-and-gas-industry]]></category>
            <category><![CDATA[leak-detection]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[methane-emissions]]></category>
            <dc:creator><![CDATA[Jeff Kimmel]]></dc:creator>
            <pubDate>Mon, 12 Dec 2022 15:27:37 GMT</pubDate>
            <atom:updated>2022-12-12T15:35:05.471Z</atom:updated>
            <content:encoded><![CDATA[<p>Reduce Methane Emissions and Cut Costs with AI and Machine Learning</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*Lc-Y6Cpb5ZFt6_nQ" /><figcaption>Photo by <a href="https://unsplash.com/@selimarda?utm_source=medium&amp;utm_medium=referral">SELİM ARDA ERYILMAZ</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>A study by the Environmental Defense Fund determined that over a 5-year period, <strong>methane leaks in the US were 60% higher than originally estimated by the EPA</strong>.</p><p>The wasted gas is enough to fuel 10 million homes for a year, <strong>lost gas that’s worth an estimated $2 billion</strong>. In addition to the economic impact, the leaked <strong>methane is 80x more harmful than CO2 </strong>during the 20 years after it is released into the atmosphere.</p><p>The resulting impact leads to devastating economic loss by both consumers and the Oil and Gas industry as a whole while also accelerating climate change.</p><h3>Leak Detection and Mitigation with AI</h3><p>The federal government, as well as the industry as a whole, is trying to do their part in reducing emissions. Successfully doing so will require a multi-prong approach. One such approach will be focusing on leak detection and repair.</p><p>With the acceleration of sensors and Industrial IoT, oil and gas companies are rich with data. This data can be combined with machine learning to detect leaks in an accurate and timely manner, allowing for quick remediation.</p><p>Technology now allows for numerous ways to detect the colorless and odorless methane gas. On the more expensive side, cameras deployed in the field or even from space can be utilized to monitor for and detect large methane leaks. In June 2022, researchers at Spain’s Polytechnic University of Valencia, said they uncovered the latest known super-emitter event at an oil and gas platform in the Gulf of Mexico.</p><p>40,000 tons of methane were discharged over a 17-day period in December 2021. Researchers said the release may never have been known to the public if not for the fact that it was captured by a European Space Agency satellite.</p><p>However, while this technology has proven effective, it is mainly accurate for only the largest of leaks and difficult and expensive to deploy at scale.</p><h3>A Machine Learning Approach</h3><p>A more scalable approach is to monitor pipelines and equipment via low cost pressure, flow, and even vibration sensors. Through machine learning, a computer can learn what normal operation of a pipeline of a machine looks like. With this trained model, systems can be monitored in real-time to look for a change in the data profile that is indicative of a problem such as a leak.</p><p>For example, simply by monitoring pressure along different points on the line, a machine learning algorithm can find patterns of normal in order to trigger alerts when the pressure or pressures go above or below a learned threshold.</p><p>More importantly, for systems that have different states such as numerous valves, machine learning can learn normal based on the state of the valves being opened to find patterns that humans and statistics would otherwise struggle with.</p><p>To learn more about how Elipsa’s no-code AI platform can detect leaks and help cut down on costs and methane emissions, <a href="https://calendly.com/d/g7fg-k67s/elipsa-predictive-maintenance-discussion">schedule a call today</a>!</p><p><em>Elipsa is a leader in delivering fully automated, scalable AI solutions for industrial IoT applications. Elipsa’s Intelligent Monitoring Platform seamlessly deploys across any workflow on the edge or in the cloud, increasing the availability and output of critical equipment. Elipsa’s self-training AI models and bolt-on approach enable AI deployments that are simple, fast, and flexible. To learn more, please visit </em><a href="https://www.elipsa.ai/"><em>https://www.elipsa.ai</em></a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=7e50f346f5b1" width="1" height="1" alt=""><hr><p><a href="https://medium.com/elipsa/ai-spotlight-leak-detection-7e50f346f5b1">AI Spotlight: Leak Detection</a> was originally published in <a href="https://medium.com/elipsa">Elipsa</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Configuring the NCD Predictive Maintenance Sensor]]></title>
            <link>https://medium.com/elipsa/configuring-the-ncd-predictive-maintenance-sensor-2d5532626524?source=rss-13d5c3943eef------2</link>
            <guid isPermaLink="false">https://medium.com/p/2d5532626524</guid>
            <category><![CDATA[sensors]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[predictive-maintenance]]></category>
            <category><![CDATA[industrial-automation]]></category>
            <category><![CDATA[iiot]]></category>
            <dc:creator><![CDATA[Jeff Kimmel]]></dc:creator>
            <pubDate>Wed, 10 Aug 2022 13:53:32 GMT</pubDate>
            <atom:updated>2022-08-10T13:53:32.924Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*EABZCdK_rsWjXjxE.jpg" /><figcaption>ncd.io</figcaption></figure><p>Over the course of the next several posts, we will detail how to take the NCD Predictive Maintenance sensor, fresh out of the box, and connect it to the Elipsa platform to start intelligently monitoring a range of equipment via Elipsa’s automated machine learning capabilities.</p><p>In the first post, we start with the hardware and walk you through how to configure the device to start generating and transmitting data to later be used by Elipsa.</p><h3>NCD Hardware</h3><p>To start, we are utilizing <a href="https://store.ncd.io/product/iot-wireless-predictive-maintenance-sensor/?attribute_would-you-like-an-external-power-supply=No%28Power+Sensor+Using+Internal+Battery%29">NCD’s Industrial IoT Wireless Predictive Maintenance Sensor V3</a>.</p><p>As the name implies, this is an industrial strength sensor capable of detecting various measurements of vibration, as well as temperature and current. These data points are utilized in driving AI-based predictive maintenance for multiple types of equipment, most notably rotating equipment.</p><p>While the sensor is capable of detecting, we also need to transmit what is detected. For this, NCD offers a variety of different modems and gateways. In our case, we are utilizing the <a href="https://store.ncd.io/product/industrial-iot-wireless-mega-modem-usb-wifi-bluetooth/">Industrial IoT Wireless MegaModem</a> as we want to take advantage of a few different connectivity options, namely Wi-Fi and MQTT as we will get into with future posts.</p><p>With the hardware in hand, it’s time to walk through the steps of setting up the equipment to start generating data.</p><h3>Configure MegaModem</h3><p>To configure the modem, we utilized the <a href="https://ncd.io/mega-modem-user-guide/">MegaModem user guide</a> on the NCD website.</p><p>First, you need to ensure the device is in setup mode. This is the initial state of the device when first powered on, but since we had a bit of trial and error in the setup process, you may need to enter setup mode manually.</p><p><strong>To enter setup mode manually:</strong></p><ul><li>remove the lid of the MegaModem, press and hold the CFG button until the LED flashes blue. <em>In our experience, the LED starts flashing green after ~5 seconds. Continue to hold the button until the LED turns blue after ~15 seconds.</em></li></ul><p>With the modem in setup mode, you can now configure the different ways to connect to the device for attaching sensors and streaming live data.</p><p><strong>Access the MegaModem Config Page:</strong></p><ol><li>Disconnect your computer from all LAN and Wi-Fi networks.</li><li>Search Wi-Fi networks on your computer and connect to the MegaModem network with password <em>NCDBeast . Note: the MegaModem network will only appear when the device is in setup mode. If you do not see the network listed, retry the manual steps above.</em></li><li>Once connected, your browser should redirect to the MegaModem config page but if not it is assessable by entering 172.217.28.1 into your web browser. <em>Note: if you do not connect from all other networks you may not be able to access this page.</em></li></ol><p>Initially, we configured the modem for USB and Bluetooth connectivity.</p><p>Looking at the resources section of the sensor’s webpage, we sought out to configure the modem and sensor using Alpha Station. Downloading and running Alpha Station, we were able to connect to our modem using Bluetooth. The software also detected the sensor but with an error ‘Unknown Data Source’.</p><p>Looking into this in the NCD community forum, we noted that our particular <strong>sensor is not compatible with Alpha Station and needs to be configured using Node-Red</strong>.</p><p>As we will note below, we also had issues connecting our modem to Node-Red via USB/Serial because of some configuration issues specific to our computer so we went back into setup mode to configure the Wi-Fi settings.</p><p><strong>Configure MegaModem for Wi-Fi Access</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*TmD-J1MftrNblV5l.png" /><figcaption>MegaModem Config Page</figcaption></figure><p>Back on the MegaModem configuration page, we clicked on the Wi-Fi section. Using the dropdowns we found our correct Wi-Fi network and entered the password for our network.</p><p>Initially we sought to utilize DHCP; however, the configuration page does not display the IP that is pulled for the modem and we did not have an easy way of finding the assigned IP on our lab network. So, we assigned a static IP instead.</p><h3>Configuring the Sensor via Node-Red</h3><p>With the modem connectivity established via Wi-Fi, we moved on to setting up the sensor.</p><p>In addition to our detailed steps below, <a href="https://www.youtube.com/watch?v=lGCIVoPd7Bs">NCD also provides a video</a> detailing how to utilize their Node-Red nodes to configure enterprise sensors.</p><p>We utilize Node-Red in other applications and thus we already have it installed but for users new to Node-Red, <a href="https://nodered.org/#get-started">download it for free from their website</a>.</p><p>To configure the sensor and connect it to the modem, you first need to download the <a href="https://flows.nodered.org/node/ncd-red-wireless">NCD Wireless nodes</a> through Node-Red and then configure them within a Node-Red flow. Below are the series of quick steps to follow:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*rbcLB0KFF9GBfroT.png" /><figcaption>Node-Red Manage Palette</figcaption></figure><p>To get started, we will first need to install the correct NCD nodes to the Node-Red palette (the list of nodes on the left hand side). To do so, click the menu icon in the top right (three lines) and select <em>Manage Palette</em>.</p><p>Click the install tab.</p><p>Where it says <em>search modules</em> enter <em>ncd-red-wireless</em> and click the install button.</p><p><strong>Note</strong>: we originally encountered errors when trying to install the nodes. After finding a <a href="https://discourse.nodered.org/t/loading-ncd-red-wireless-on-windows/43896">related issue in the NCD community</a>, we needed to downgrade our version of node.js to 12.20.1. We were able to find and <a href="https://nodejs.org/download/release/v12.21.0/">download node.js 12.20.0 here</a> and install the nodes without issue.</p><p>You will now see the NCD Wireless Gateway and Wireless Device nodes in your palette.</p><h4>Configure the Modem Node</h4><p>First, we need to attach to our MegaModem by dragging a <em>Wireless Gateway Node</em> onto the wiresheet on the right.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*m3hfCKbdjCiqUrb-.png" /><figcaption>Add Wireless Gateway Node</figcaption></figure><p>When the Properties details appear:</p><ol><li>Name your node</li><li>Click the edit icon next to <em>Serial Device</em>. When the second Properties page appears</li><li>Name your Modem</li><li>Select TCP (Ethernet/Network) from the dropdown menu</li><li>Enter the IP address of your modem that you configured earlier</li></ol><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*o0SK_VDPg8B9PCSW.png" /><figcaption>Add MegaModem</figcaption></figure><p>Your modem is now connected to Node-Red.</p><p>To confirm this, drop in a <em>debug </em>node and connect it to the output of your <em>Wireless Gateway</em>.</p><p>The result is that when your modem/gateway connects to Node-Red, the debug log on the right will show the MAC address of your modem which will match the sticker on the side of your MegaModem case.</p><p>Your modem is now connected and ready for the next step of configuring your sensor to send data through it.</p><h4>Configure Your Predictive Maintenance Sensor</h4><p>First connect the vibration and current/temp sensors to ports on your sensor case. Remove the cover and switch on the sensor as they are shipped off by default.</p><p>To turn on the sensor, find the black switch located next to where the batteries from the case’s top connects to the main board. Push the switch towards the wall of the case to turn it on.</p><p><strong>Configure Device in Node-Red:</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*-5IgI86u-uBZBdEg.png" /><figcaption>NCD Sensor Configuration</figcaption></figure><ol><li>Drag an NCD Device node onto your wiresheet and name the node</li><li>On the properties form, select the modem name that you just added in the <em>Serial Device</em> and <em>Serial Device for Config</em> drop downs.</li><li>Double click the search icon to the right of <em>MAC Address</em>. This will show you the MAC address of all sensors that the MegaModem detects. Find and click the MAC that matches the value on the sticker located on the side of your sensor’s case.</li><li>The sensor type should automatically populate with ‘Condition Based/Predictive Maintenance’</li><li>Check the box next to <em>Auto Config</em></li></ol><p>Your device now is setup to connect to your modem with the default settings. For our particular use case we changed the following settings:</p><ol><li>Check the <em>Active Box</em> under <em>Node ID and Delay</em> and change the delay to 1 (1 second). This will set the sensor to record data approximately every 1 second.</li><li>Under <em>Measurement Mode</em> check off <em>Active </em>and select ‘Processed data, acc in mg…’ from the drop down. This value will change based on the use case and will change in future posts depending on the machine we are monitoring and how often we want to analyze data.</li><li>Under <em>Sampling Interval</em> check <em>Active </em>and change the value to <em>1 Minute</em>. This will collect data based on the delay setting you set in #1 and then transmit these results every 1 minute.</li><li>Under <em>Full Scale Range</em>, check <em>Active </em>and choose your preferred setting. Our test equipment does not have a strong vibration profile so we are using a lower range.</li><li>Under <em>Configure Enabled Axes</em>, select all three X, Y, Z</li><li>Click Done</li></ol><p>These settings will read vibration measurements every second calculating a variety of vibration calculations and transmitting them every minute.</p><h4>Push Config to Device/Modem</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*6PNs2lgfcmv2GEsF.png" /></figure><p>With the modem in configuration mode, take the sensor and manually do the following to pull the new configuration:</p><ol><li>Remove the lid</li><li>Push and release the Reset button</li><li>Push and hold the CFG button for ~ 8 seconds</li></ol><p>When you release the button, you will see the status of your device change in node red. It will go through a series of steps to pull the config.</p><p>When the status displays as <em>Config Complete</em>, we complete the task by taking the modem and sensor out of configuration/setup mode:</p><ul><li><strong>Modem</strong>: in Node-Red push the button on the left of the Wireless Gateway node</li><li><strong>Sensor</strong>: on the physical sensor, push the reset button</li></ul><p>Your sensor is now configured and sending data to Node-Red. To confirm this, with the modem/gateway connected to the debug node, you should now see streaming data appear in the debug log every minute as the sensor transmits data.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*M3_C1pm96w_HAbpx.png" /></figure><h3>Next Steps</h3><p>With our sensor configured and streaming data, there are a range of tasks that you can perform within Node-Red.</p><p>In our next post, we will detail how to configure the Elipsa’s AI-based Predictive Maintenance solution to integrate directly into your Node-Red instance via REST APIs.</p><p>While it appears that Node-Red is required for configuration of the NCD Predictive Maintenance sensor, many of you may not utilize Node-Red in your production application. As a result, our third post will go back into the MegaModem config to show you how to integrate your NCD sensor data directly into Elipsa via MQTT.</p><p><em>Elipsa is a leader in delivering fully automated, scalable AI solutions for industrial IoT applications. Elipsa’s AI-based, Predictive Maintenance seamlessly deploys across any workflow on the edge or in the cloud, increasing the availability and output of critical equipment. Elipsa’s self-training AI models and bolt-on approach enable AI deployments that are simple, fast, and flexible. To learn more, please visit </em><a href="https://www.elipsa.ai/"><em>https://www.elipsa.ai</em></a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=2d5532626524" width="1" height="1" alt=""><hr><p><a href="https://medium.com/elipsa/configuring-the-ncd-predictive-maintenance-sensor-2d5532626524">Configuring the NCD Predictive Maintenance Sensor</a> was originally published in <a href="https://medium.com/elipsa">Elipsa</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Next Top IoT Use Case]]></title>
            <link>https://medium.com/elipsa/the-next-top-iot-use-case-872e1125e631?source=rss-13d5c3943eef------2</link>
            <guid isPermaLink="false">https://medium.com/p/872e1125e631</guid>
            <category><![CDATA[industrial-automation]]></category>
            <category><![CDATA[iot]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[predictive-maintenance]]></category>
            <dc:creator><![CDATA[Jeff Kimmel]]></dc:creator>
            <pubDate>Wed, 03 Aug 2022 13:38:46 GMT</pubDate>
            <atom:updated>2022-08-03T13:38:46.088Z</atom:updated>
            <content:encoded><![CDATA[<p>The Internet of Things (IoT) continues on it’s upward trajectory, leading to an ever increasing number of connected devices and data points being generated.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*yGalZ-skCkWEnxc3Uot8iw.png" /><figcaption><a href="http://www.iotanalytics.com/">www.iotanalytics.com</a></figcaption></figure><p>According to IoT Analytics, two of the top 3 use cases for IoT utilizes sensors and data to connect and monitor remote assets. Equipment previously only accessible by traveling physically to them can now be monitored in a central location thanks to the growth of IoT.</p><p>Further down the list is the use case of Predictive Maintenance (PdM). In many ways, PdM has become a buzz word and has taken on different meanings with a focus on reducing the cost of equipment maintenance and downtime.</p><p>To date, PdM has largely focused on monitoring machinery within manufacturing, and to a lesser extent heavy equipment within areas such as oil + gas, aviation, and rail.</p><p>However, as those areas mature in terms of new connected devices. IoT growth as a whole is still accelerating.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/886/1*ybVoQ5InW5kELhW0BaUmew.png" /><figcaption><a href="http://www.iotanalytics.com/">www.iotanalytics.com</a></figcaption></figure><h3>Predictive Maintenance for Remote Assets/Equipment</h3><p>Looking at the expected increase in IoT connections, and coupling that with the top use cases, you can logically infer that <strong><em>the largest area of future growth will be in connecting old and new remote equipment.</em></strong></p><p>So, this raises the question, is predictive maintenance for remote assets the next key use case and untapped market for IoT.</p><h4>Remote Assets</h4><p>First off, what is included in the definition of a remote asset? Remote assets can be assets on the move such as trucks for logistics and supply chain companies, or geographically dispersed such as water pumps located in a remote field, but it could also just mean equipment that is difficult to physically access or that’s not readily in front of you to easily monitor such as an HVAC unit located on a roof. IoT is primed to connect these devices, generating data that AI is then primed to analyze.</p><h4>AI-based Intelligent Monitoring</h4><p>While AI-based PdM solutions offer the promise of being able to intelligently monitor remote equipment, a key challenge exists.</p><p><strong>Current Challenge</strong></p><p>Predictive Maintenance solutions that were early to the game were built by teams of data scientists, often manually, designed to be point solutions for industries noted above such as manufacturing. They are built for very specific machines. In addition, their AI-based approaches tend to require large amounts of historical data.</p><p><strong>Focus on Simple, Fast, and Flexible</strong></p><p>As you can see from the chart earlier, billions of devices are yet to be turned on meaning they have not yet generated a single data point of historical data. As a result, in order to turn PdM of remote assets into reality, we need a solution that is flexible enough to adjust to various types of equipment. It needs to be simple enough to use to puts the functionality in the operators hands without needing a team of data scientists to monitor a new machine. Finally, the AI approach needs to start with anomaly detection. Flip the problem on it’s head and instead of learning to find patterns of problems instead learn normal to be able to then identify abnormal. This approach requires far less data and lead’s to actionable insights in days not months or years.</p><h3>Conclusion</h3><p>The top IoT use cases show us where the value is today for the end user. Logically, to provide the next level of incremental value we can look across the existing use cases that have been proven and see which of them complement each other in a way that if combined they provide a new form of value. Predictive Maintenance and Remote Asset Monitoring do just that. So, is that the next big thing in IoT?</p><p>That appears to be where the market is going but like many innovations, road blocks exist. Elipsa is focused on providing plug-and-play predictive solutions to help turn the promise of this next way of innovation into reality to continue to provide higher levels of value to the Industrial IoT community.</p><p>Elipsa is a leader in delivering fully automated, scalable AI solutions for industrial IoT applications. Elipsa’s AI-based, Predictive Maintenance seamlessly deploys across any workflow on the edge or in the cloud, increasing the availability and output of critical equipment. Elipsa’s self-training AI models and bolt-on approach enable AI deployments that are simple, fast, and flexible. To learn more, please visit <a href="https://www.elipsa.ai/">https://www.elipsa.ai</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=872e1125e631" width="1" height="1" alt=""><hr><p><a href="https://medium.com/elipsa/the-next-top-iot-use-case-872e1125e631">The Next Top IoT Use Case</a> was originally published in <a href="https://medium.com/elipsa">Elipsa</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
    </channel>
</rss>