In the February and April issues we explored how the pruning work load expands. Toward the close of the February article, I stated “Good and properly funded VM programs experience few if any grow-in outages (0-2% of tree-related outages).” Statistics from broadly geographically dispersed utilities suggest 85% to 98% of tree-caused outages arise from some type of tree failure.

If we are to manage tree-related service interruptions, we need to examine tree failures. That starts by aligning processes to provide insight. While pretty well all utilities track outages caused by trees, in my experience there is a broad range in the utility of the data collected. Some utilities collect data categorizing tree-related outages into preventable or non-preventable. Usually only grow-ins are considered preventable. The best tracking I’ve encountered reports whether the tree-caused interruption arose from a tree within or outside the maintained right of way; whether the outage was due to a tree failure with further information about the type of failure. It may have been due to a branch failure, a trunk failure or an uprooted tree. When you add to this reporting of the serviceman and a follow-up inspection by an arborist, you begin to add detail on the tree species and over time gain insight into particular species vulnerabilities. This data can then be used to focus the hazard tree program. You may for example, find that a tree species that is only a minor component of the utility forest has a particular failure vulnerability that is often realized and, consequently, direct a more thorough risk assessment on stands of that species.

Here are some of the data garnered by a combination of servicemen and arborists reporting on tree-caused outage events.

  • Outage data: duration, number of customers affected; customer minutes lost due to tree-related outages; percent of tree-related outages of all unplanned outages
  • The percent of grow-in vs fall-in outages
  • The percent of branch failure outages
  • The percent of trunk failures and uprooted trees
  • The tree species
  • The distance of the failed tree from the nearest conductor
  • The percent of hazard trees vs seemingly structurally sound healthy trees
  • The weather conditions at the time of the interruption

Collecting this data takes time and costs money. So what is the payoff? How can we use this data?

Grow-ins occur from either underneath or adjacent to the conductors. If you can show your grow-in outages are below 5% of all tree-related service interruptions, you are demonstrating that you are properly funding the pruning program and generally, managing the trees within the right of way and directly adjacent to it. It would make it clear that 95% of your tree-caused outages arise from trees located on private property where your rights are limited.

I’m sure many of you have been in the position of wanting to remove branches overhanging distribution lines. There’s a considerable probability that your initiative was not well received by the community. People like their streets with big tree canopies and for some reason, do not find the blue sky you open up by removing all overhangs to be a positive aesthetic. Do you think you might have a better chance with the community if you had data indicating that over 60% of your tree-related interruptions were due to the failure of branches that were overhanging conductors?

At least the regulator would be aware of what you propose to do to improve reliability and why. Add that it is not uncommon for a major ice storm to cause crown damage in over 10% of the trees. Studies of tree damage after the 1998 ice storm in the northeast found 40 to 50% of the trees had damaged crowns (http://www.dec.ny.gov/docs/lands_forests_pdf/ice99.pdf). That should create a pretty clear picture of what will happen to the electric system that tolerates tree branches overhanging conductors when you experience an ice storm or wet snow loading. That’s not to say removing overhangs will eliminate all outages arising from branch failures, which could still occur through wind throw, but the probability is greatly reduced. It’s difficult to gain both regulatory and community support without specifics on the risks and benefits.

I have often heard from utility arborists after a major storm that of the trees that caused an interruption, their guestimate was that over 90% would have been considered healthy rather than hazard trees. This does not seem to raise sympathy with regulators, politicians and I would imagine, the public. If, however, because you have arborists investigate tree-caused outages, you are able to definitively say as Puget Sound Energy and National Grid Transmission have stated, that 68% of tree-caused outages arose from trees that were not hazard trees, you have created a baseline and context for major storm damage. Would it be useful to have your regulator understand that your VM program can realistically only improve upon the other 32% of tree-related outages? Certainly it makes sense that if under normal operating conditions the majority of outages arise from apparently healthy, structurally sound trees that this percentage would be even higher during major storm events. Healthy, structurally sound trees are not targeted by the VM program. In that limited sense, outages arising from healthy, structurally sound trees are non-preventable.

The distance of failed trees from conductors is useful data. First this informs the vegetation manager whether crews are doing a good job of identifying hazard trees along the right of way edge. If they’re not, it can be rectified by more funding, training or a more rigorous methodology. However, if the failed hazard trees are well beyond the forest edge, the challenge to reduce tree-related interruptions is far more complex. Identifying tree hazards inside a forest stand is far more difficult simply through limited visibility and consequently, it is far more costly. The data is the first step in helping the regulator to understand that you are trying to maximize the public good. It is an area where increased spending may or may not provide a commensurate return in improved reliability.

The intent here has been to illustrate that there are two benefits to collecting detailed data regarding tree-related outages. The data can be used to improve the VM program. Equally important, however, is that the data facilitates the communication of realistic expectations for system performance and improvement.