This use case describes how ‘Edge Computing’ can apply not just to cater very high end, large infrastructure and complex processing applications but also very useful for small to medium scale applications as widely in demand by today’s fast growing industry world. With ‘Pump House Monitoring’ application example, it demonstrates various advantages the water utility sector can gain over conventional solutions.
Water Pump House / Pumping Stations are mechanism that can be used to transport water from one site to another usually from its source to consumer e.g. supply water to canals, circulate water in treatment systems, supply water to residential areas etc. Few important parameters like pump motor input power supply, its power consumption, water flow, water pressure, time scheduled or demand based pump on & off control etc. are to be taken care for its smooth functioning so that to meet demand-supply in efficient manner. Again, this can be remote site set-up so all such monitoring & control need to be done remotely from utility’s back office (control station) over wireless / wired network connectivity.
Conventional solution utilizes below components -RTU or PLC
- Cellular Modem or Ethernet for remote connectivity.
- RS 232/RS 485 serial ports, analog inputs and outputs for energy meters, flow meters, pressure sensors etc.
- Digital inputs and outputs for pump on-off control and basic security e.g. panel door open/close, power fail or any overload trip indication etc.
- Limited memory for various event log/history data storage.
This is how Conventional Solution manages basic application. All required sensors / peripheral devices are wired to RTU / PLC, data logged into local memory, data processing, data sent to utility control station over cellular or Ethernet connectivity. It can log and generate power fail or any pre-set threshold limit alert messages as well for supply voltage fluctuation, overload trip etc. Store history data during connectivity loss. For any maintenance activity or fault recovery, operator needs to visit site and take corrective actions.
Challenges/Limitations with Conventional Solution
Though it can manage basic application in simple manner, it has few limitations as well.
High Network dependency to get real time alerts. This can be critical if user needs to take immediate corrective action in response and hence may increase system downtime.
Operator visit becomes must for any system level diagnosis & maintenance activity in case efficiency loss reported. Such is not always favorable for the far remote sites or if located at high altitude, hilly area etc.
Wired connectivity with limited interface options for all peripherals installed inside or outside pump house premise that can lead to reliability issues.
No premise security, so theft threat. If needs to build this feature, requires additional system solution with IP Cameras & dedicated server including integration efforts to user control room in real-time manner.
Lack of preventive maintenance mechanism leads to high maintenance cost/failure recovery. To have this feature added in conventional system, need to add specific sensors along with pump, have added interface channels at RTU/PLC end, data capturing, logging and transporting mechanism which again demands for high end RTU/PLC solution, reliable connectivity for large data transport, software solutions/cloud components with data analytics for preventive maintenance. All these means more dependencies and cost addition.
Edge Computing Solution and it’s Components
Edge Computing is computing that’s done at or near the source of data, instead of relying on the cloud/server to do all complex analytics work. With this it is more efficient to cater data latency, bandwidth and security related issues hence attractive feature to have for any real time monitoring & control application solution. To have this feature, it demands for certain capabilities in control system solution depending on complexity of tasks involved with targeted application.
Leveraging this for pump house monitoring application, here we list out some of key features/components example to integrate with device solution or say build ‘Edge Computing Platform’
Required Components –
- High speed (800 MHz~1 GHz) Processor
- Linux Operating System -2GB RAM, 4GB Flash
- Video Processing Unit (VPU), Image Processing Unit (IPU), 2D/3D Graphics Processing Unit (GPU), Asynchronous Sample Rate Converter (ASRC) etc. hardware accelerators
- Pre-trained data analytics models for Motion Detection, Human Detection and Pump Health status monitoring etc. The models will be trained with ML frameworks, i.g. Tensorflow, Keras, Pytorch or Mxnet on a powerful system such as a Deep Learning server. Then the model that is trained will be exported to the edge device.
- Open-CV or Pillow library for IP Camera initialization and Image/Video Pre-Processing
- Cellular Modem, Ethernet Port, POE+ Port, RS 232/RS 485 Port
- Analog and Digital inputs & outputs
- Zigbee / Bluetooth wireless interface options
- HART Interface option
Work Flow & Advantages with Edge Computing
Diagram here shows how edge computing platform can replace any conventional system solution.
- It manages to read all those same parameters from various field sensors but with added option of wired and wireless connected sensors, e.g. Pump
- Vibration Sensor, Water Leak Detector, IP Camera, HART Sensors etc.
- After various data acquisition, platform processes various input parameters and pass required data to respective AI (Data Analytics) tool running there on Edge Platform.
- With conclusive data output from all required data analytics, it controls back the application process as and when needed in real time
- Generates if any alert on potential pump failure, water leakage, security concern etc. observed & report to control station. Also takes corrective action as Pre-programmed.
- Log all events and report to control station as well as maintain history data for defined time duration.
- No critical dependency on network availability/control station connectivity as Edge Platform itself log, process, identify & take responsive actions in real time.
- With powerful core inside, user can fully access/control the process remotely for any configuration updates or remote debugging.
- Powerful core offers added wireless connectivity options for peripheral field sensors like Zigbee/Bluetooth which can solve any reliability issues in long wiring outside pump house premise.
- Built in feature like HART not just monitors but also controls the configuration of HART based smart field sensors. Its purpose it to avoid human dependency incase any configuration change/calibration etc. required of filed sensor during its filed operation.
- POE+ Port offers Power + Interface to IP Cameras. With image/video data analytics tools at Edge Platform, this can monitor premise security for any theft threat or un-authorized access.
- With various pump operation parameters Data analytics at edge, it can predict pump health status. This saves lot of failure recovery cost by alerting maintenance team on potential pump failure situation in advance.
- Reduced network dependency, avoid data latency and data bandwidth requirement.
- Full access/device control remotely, can avoid operator’s site visit and reduce maintenance cost.
- Feasible to connect and monitor wireless sensors, smart field sensors, IP Cameras. with benefits out from all such added features, it justifies the overall system solution cost.
- Predictive maintenance saves part failure recovery cost and reduce system downtime.
- Better Data privacy/security as all critical field data remains at edge device itself.
- Single device solution saves overall system cost, cloud components cost, integration efforts & time.
With Edge Computing and Analytics tools implemented with Edge Platform, data processing can be faster, responsive action can be taken in real time, overall system monitoring and data analysis for end user becomes simpler as conclusive notifications they receive in alert form. All achieved with minimal system components that enhances system reliability and optimizes solution cost.