Learnings

Outcomes

We completed testing of two different prototype link box sensors in early 2019:

Option 1: a simple link box temperature sensor

Option 2: a more sophisticated sensor which measured temperature, moisture and current.

Both devices met the testing requirements. After further internal engagement, it was decided to proceed with developing a simple link box temperature sensor. We reached out to the supplier of Option 2 prototype and asked if they wished to develop a prototype which fit the Option 1 specifications. They declined because the simple device design did not align well with their overall business strategy and key focus areas.

The Option 1 prototype was taken through to a live trial.

We received 50 sensors later in 2019 and completed network installation by January 2020. The sensors stayed in situ until Q4 2020.  The configurable sensors were set to measure temperature every 15 minutes. This built up a database of actual temperatures experienced at the bell cover of link boxes. The data from the sensors was accessible through a web-based platform. The project observed daily sinusoidal temperature variations, which we attribute to network loading and daily temperature variation.  

During the trial, the sensors raised alarms and alerts when the configurable limits were met. The alarms came through to a dedicated web-ased software tool, as well as via email to staff. Throughout the trial, the alarm and alert temperatures were manually altered with a view to identify what an appropriate level might be. In the vast majority of cases, alarm limits were hit during periods of very hot weather. This was not the desired outcome, as spurious alarms during warm weather meant the sensors were not able to accurately identify link box faults. 

We learned that a static alarm limit was not appropriate. However, by reviewing the measured temperature data, some events could be observed from the link box sensor data:

1.  There was one network fault, which was identified by a customer calling in about a power outage. At the link box right outside the property, an abnormal temperature drop could be seen from the data. We attribute this to a loss in network load through the link box. 
2.  There was another event where, although the link box sensor alarm limit was not breached, an abnormal rise in temperature was manually observed. An engineer was dispatched to site, who was able to confirm that the temperature underneath the bell cover matched the sensor, and that the link box was faulty. The box was subsequently replaced. This was encouraging, as it demonstrated that temperature can be used to identify link box faults. However, a more sophisticated alarm limit strategy would be required.
 

The key project learning was that a more sophisticated alarm system would be required to make this technology valuable to the wider business. It is not sustainable to rely on manual data analysis as per these two examples.  

Another key project finding was that static temperature settings are not appropriate. This is discussed further in the Lessons Learned section. 

Lessons Learnt

Key Lesson 1 – Use Cases: There are multiple use cases for a link box sensor, depending on whether or not the aim is to protect the link box or get information about the surrounding connected network. The use cases identified in this project included the identification of:

-          Overheating in a link box;

-          Presence of smoke in a link box;

-          Any other fault with a link box;

-          Flooding in a link box;

-          A nearby cable fault;

-          Direction of travel of a recent nearby cable fault;

-          Distance to a cable fault;

-          Tampering with the link box;

-          Being able to gather this information from a sensor without lifting the bell cover; and

-          Gathering information from a sensor without attending site.

These use cases increased the time required to engage with stakeholders, agree requirements and procure the device. The project attempted to rationalise these use cases and identify the best combination of sensor features to be useful to distribution network operation. We learned from the market that the TRL of the technology available for specific features varied greatly. For example, the TRL of a temperature sensor is much higher than the TRL for communications technology for communicating from within the link box.  We also learned that only some use cases were suitable for a device housed above the bell cover. Current measurement must take place below the bell cover because access to the electrified components of the link box is required. These are protected beneath the bell cover.

Therefore, it was suggested that individual functionalities of link box sensors were prioritised and perfected before being combined. More research and development is required for a user-friendly device to be produced. Due to the large population of link boxes across DNOs, any device installed must not significantly increase the time it takes to inspect or otherwise interact with a link box. 

In the design of new monitoring equipment, it is important to have a specific use case in mind and to engage with the relevant internal stakeholders to ensure that these are needed and relevant. These learnings are benefitting other UK Power Networks projects such as the Active Response NIC project. 

Key Lesson 2 – Moisture Sensing: The presence of water in a link box can be a cause of failure, especially if a box becomes flooded. Flooding in a link box can cause a short circuit which is a potentially dangerous situation. Therefore, the project hypothesis was that sensing moisture in a link box could lead to network benefits. However, through the project, we learned that a sensor can be used to monitor the water level, or a different type of sensor can be used to monitor relative humidity.

Upon closer consideration, we determined that neither of these sensors would add significant value to a link box sensor. In the case of relative humidity, it is not expected that a high relative humidity alarm would be a credible risk to the link box. This is because some humidity and moisture is to be expected in the link box environment by default. In the case of water level, in order to give useful information, this would have to be placed in multiple places on the bottom of the link box (below the links). Otherwise a water level alarm would not activate early enough to send someone to site to rectify the problem before significant damage is encountered. It is impractical to access this section of a link box, and water level sensor technology is too big to install at the bottom of a link box. 

Key Lesson 3 – Bell cover temperature as proxy:  A key hypothesis at the beginning of the project was that measuring the temperature on the outside of a bell cover would be an acceptable proxy for temperature in a link box (with an average difference between the two). A key project learning has been that this is not an acceptable proxy. There appear to be too many variables influencing the temperature on the outside of the bell cover, but that the ambient temperature and electrical load both have a large influence. Because a link box has a large thermal mass, there appears to be a lag on the measured temperature. More work is needed to understand the relationship between temperature inside and outside the bell cover, the associated lag and impact of varying load and ambient temperature. 

Key Lesson 4 – 4G comms is sufficient if the device is located above the bell cover. This project has also benefitted from the lessons learnt from the project NIA_SPEN_0026 Linkbox Monitoring using Narrow Band IoT.