League Open Forum

Expand all | Collapse all

Real-time Computing Embedded into Sensing and Controls

  • 1.  Real-time Computing Embedded into Sensing and Controls

    Posted 04-21-2020 16:51
    Good day. My name is Chris Doerfler. I am a cofounder in 3DFS Software-Defined Electricity and SAM Controllers. These companies leverage a new real-time computing technology with unlimited parallel computing capability that has been applied to sensors, with a particular focus of sensing electricity. The advantage provided is full digitization of the analog electrical signal creating what is essentially an electricity microscope, opening up a new world of visibility into electricity and along with it, a precision way to interact with it.

    Using this technology, machine control systems have precision, real-time feedback using the electrical signatures which are absolute and cannot be altered or mimicked, allowing the control decisions to be made with orders of magnitude more information and analysis at significantly faster speeds. 

    Stated simply, a machine requires energy to operate. That energy comes in the form of electricity. The pattern which the machine consumes electricity is unique, just like a thumbprint, because every one of the machine's internal components is slightly different (wire thickness varies, electrical tolerances vary, etc.). Fully and continuously detecting this pattern provides real-time information on the machine that can be used to make better decisions on machine performance. 

    In other words, machines universally perform more consistently with higher output and reduced energy consumption.

    I am opening up this discussion which I hope to be collaborative and educational. My goal is to explain this new method of sensing and machine control to those interested in learning more. There is quite a bit of innovation within this technology so I encourage everybody to ask questions and I will be happy to answer them. 

    We have already applied this method to air compressor pumps, water pumps, batteries, power electronics, generators, transformers, inverters, liquid levels, and many more industries.

    Some great questions can be formed about these applications. Other areas of interest are about the computing method, data acquisition method, modeling capabilities, artificial intelligence algorithms, processing power, and so much more. 

    Thank you for your interest and participation. Please ask away.

    ------------------------------
    Chris Doerfler, Cofounder 3DFS, SAM Controllers
    ------------------------------


  • 2.  RE: Real-time Computing Embedded into Sensing and Controls

    Posted 04-22-2020 13:20
    Last year I purchased a Sense https://sense.com/ device.  It was a little too far away from my wifi to be very informative but otherwise did what I had expected.  One thing I wish it had done was detect surges that were occurring when my A/C compressor started.  That could have saved me some costly repairs.

    What other applications do you see for this kind of monitoring?

    ------------------------------
    Aaron Davis
    ------------------------------



  • 3.  RE: Real-time Computing Embedded into Sensing and Controls

    Posted 04-22-2020 18:53
    Great question. The application that you are highlighting, residential device detection, is certainly an application that can be pursued. This is where the combination of electricity analytics and computing power become a very powerful combination. There is no limit to the signature that can be detected.

    With electricity analytics, some major performance improvements will be precise device identification without the need for "learning the signature" and also differentiating between identical loads (same brand and model) because differences between them can be detected. 

    Ultimately, non-intrusive load monitoring is nice, but the goal is predictive analytics. Once the all signal, no noise baseline electrical signature is identified, the next step is mapping the change in that signature over time. Any change in that signature essentially translates to wear and tear of the internal components over time. These internal components are all approximately the same types of components (i.e. resistors, capacitors, etc.) with similar degradation curves, and so on. 

    All of this information can be accessed now for every single device that consumes electricity, leading to a world with 100% non-intrusive load monitoring with precision predictive analytics. Instead of plugging into a power grid, the future will be plugging into an information and power grid. 

    There are even further applications at the individual electronics level, for example control systems for machines, which has significant carbon emission reduction benefits. 



    ------------------------------
    chris doerfler
    ------------------------------



  • 4.  RE: Real-time Computing Embedded into Sensing and Controls

    Posted 04-23-2020 13:39
    The ability to balance and monitor power is very useful to the built environment in addition to manufacturing.  I'd like to find an affordable answer to the ability to monitor and control electricity flow.  I see a number of devices in the $1500 range, but with installation that's prohibitive for most home owners after installation costs.  Is there a lower cost approach to the equipment and installation?  Where should I look?


    ------------------------------
    Michael Linn
    Dallas
    ------------------------------



  • 5.  RE: Real-time Computing Embedded into Sensing and Controls

    Posted 04-22-2020 14:29
    Chris,

    What you describe parallels very closely what we in the electric utility industry (I'm with Oncor, a transmission/distribution utility in the ERCOT Grid) do with data from, as you mentioned, power electronics (our own switchgear, etc.), but, increasingly, from the 3.5 million now-digital meters ​at consumer sites.  With that data, we are finding a treasure trove of diagnostics and prognostics data not just about the site (which a TDU only needs in the case of anomalies or outages), but also about our utility-side-of-the-meter infrastructure.  Taking care of equipment failure before the fact is obviously preferred, both by us and our consumers

    Optimization, vs. just prognostics and diagnostics, would be an obvious next step.  So, some questions to you and additional food for thought. 

    The signature analysis you describe is at device levels.  What is the nature of the installed sensor (in-line with power, busbox, proximal plug-in, inductive, …?), and what sort of sampling rate and power use/quality indicies are you talking about?  Note that, at aggregate levels, many of these 'device' signatures can be pulled from even the whole-home meter data (and not just by the Utility, via open protocols to which we by regulation must adhere), also noting that a number of firms are selling 'Smart Home' devices that are installed at breaker boxes, or on circuits, or at specific devices (Pecan Street in Austin has an ongoing DOE project to this end) and do not only the type of signature analysis that I think you're talking about, but also in some cases power control. 

    My next question would be whether you've looked at efficiencies not specifically at the per-device level (e.g., motors/pumps, as you've mentioned, being a great target), but at system levels, such as co-varying volt/var influences across all devices on a circuit to attain either an expected 1.0 power factor, or a specific target that may be a useful momentary offset for a second, connected system (e.g., the Grid)?

    If I was posting this question to my other EarthxLeague profile (my Oncor profile), I would more closely couch any statement regarding the vast efficiencies that COULD be attained in power delivery if loads were not just viewed as passive devices to be 'fed'.  As more homes and businesses bring to the 'table' their smart inverters (in their EVs, on their solar panels, or in home battery systems), or even just demand-responsive dumb loads (via Nest theremostats or even just Alexa chatting with the home refrigerator), the opportunity to connect the dots for these higher efficiencies, without just adding capital equipment to the Utility side of the system, becomes possible.

    ------------------------------
    Randy Boys
    ------------------------------



  • 6.  RE: Real-time Computing Embedded into Sensing and Controls

    Posted 04-22-2020 18:53
    There is a lot to unpack there, so I will take each point individually.

    Regarding the 3.5 million meters worth of data, there is certainly some valuable information that can be gleaned from that information. Imagine that computing power and storage were not an issue and you could have analytics at GHz frequencies what could be revealed. This is essentially what we do.

    Regarding the signature analysis, the analysis is through flex coil current transformers, one on each phase. The sampling rate is tuneable and depends on what you are trying to measure, but for our most advanced application there are 26 channels synchronized receiving 6ns input at 24bit resolution. The advantage of this real-time parallel computing is that the current, voltage, power quality, harmonics, reactive power, etc. can all be calculated and acted upon at the same time. The more accurate the data acquisition, the more accurate the interaction can be with the signal.

    Regarding volt/var influences. Yes. This is the most advanced application of Task Oriented Optimal Computing ("T2C"), to control power electronics and optimize the flow of power in real-time. We created the VectorQ Series power controller which is installed in a panel and matches the impedance for all loads in that panel simultaneously. The results in savings ranges from capacitive loads like electronics savings 10-15% to inductive loads like motors and pumps savings 20-25%. In particular, three phase motors have demonstrated savings up to 60%. There is a youtube channel on Software-Defined Electricity that exhibits this functionality.

    Regarding the efficiencies that can be gained, the heart of efficiency in electricity control lies in data and analytics. The heart of useful data and analytics lies in speed and accuracy. Real-Time, parallel computing is the only way to achieve these constraints. Moreover, when it comes to devices like inverters and transformers, the present methods cause a lot of instability during operation, just like all of the switch based power electronics that exist. In the future, these devices will include real-time power correction and integrate seamlessly with the flow of power. There is no need for these devices to decrease the stability of the grid and increase the carbon footprint of electricity transfer.

    ------------------------------
    chris doerfler
    ------------------------------



  • 7.  RE: Real-time Computing Embedded into Sensing and Controls

    Posted 04-23-2020 13:39
    Can this tech self-identify equipment in a home accurately (avoiding complex and expensive set-up).  Can it detect the difference between a dryer motor and a similarly rated moter?

    ------------------------------
    Michael Linn
    Dallas
    ------------------------------



  • 8.  RE: Real-time Computing Embedded into Sensing and Controls

    Posted 04-23-2020 14:55
    In the most advanced application of Software-Defined Electricity, the function is to match centralized power supply with distributed power demand in real-time. The way in which this is done is by instantly identifying, analyzing and tracking every electrical signature and all the variations within continuously. The only way to perfectly match supply and demand is to know exactly what the supply and demand is.

    While the automatic identification of signatures is great, this is different than a person wanting to track a signature. Here, some work is required. The system needs to know which signature you would like to track, so the owner would have to teach the system by turning the device on or off. This alerts the system as to the signature that must be tracked.

    Once a device signature has been registered in the system, it will pop up as a suggestion if the owner attempts to register another of the same device (e.g. same brand and model).

    Yes, all motors, every single one has a different electrical signature. There is no way to wind a motor identically with another. There are too many variances.

    ------------------------------
    chris doerfler
    ------------------------------