Effectively planning for your personnel requirements at an airport can be a real challenge. First and foremost, we have to "take them as they come" with flight schedules dictated by airlines* and this means peaks and troughs in demand that are difficult to match with people.
In mid-2016, the Queenstown Airport was moving to extended hours operations. Previously, the airport had been limited to daylight operations and could be covered with a single shift of people each day. With after-dark flights being approved, the airport was moving to a 18-hour-plus day but it wasn't as simple as doubling the size of the team.
For the first season, scheduled after-dark flights were minimal but daylight flights were also continuing to grow. The challenge for me was to identify my personnel needs using some foundation that would not only support my request for additional people but also support future requests as the airport continued to grow.
Not Quite Starting from Scratch
While it seemed like I was starting with nothing, it became clear to me quite quickly that the organisation had already thought a lot about its needs but had never put it into a coherent model. When speaking to floor staff and duty managers, they used words like heaving, stretched, tight, blocked, etc. They knew what was going wrong as well as where and when they needed help.
What struck me was the organic qualities of the comments and it conjured up an image of the airport as a living creature that would suffer pain as passengers passed through it. I was inspired to develop a model that would predict when and where the airport would feel that pain.
Queenstown Airport has all the standard airport pinch points - check-in, security, customs. It also has a couple of rather novel ones such as on the apron where flights of different categories can't mix and require "traffic" management.
Again, the team knew where these issues were and helped me to understand what impact each area had and how they could assist in making it better.
Charting Passenger Flow
Not all passengers behave the same way and not all flights have the same requirements on their passengers. Only international passengers have to be processed by customs and biosecurity officers and only departing passengers pass through security, etc.
So, in order to calculate the impact of an individual flight, I had to vary the model depending on whether in was domestic-v-international and arriving-v-departing. Within these four categories I then needed to establish the timings for passengers at each stage of their journey and make estimates of how many would reach each stage at approximately the same time.
My model was quite basic and but it was modelled down to five minute intervals across seven pinch points.
I wish I could say that this number crunching was done using some sophisticated, high-tech software platform. I can't. It was all done in Excel using filters, look ups and a host of other formula to produce what I called the "Pain-o-meter" (Trademark pending**).
The result was an extremely noisy look at when the airport would be under pressure. At five minute intervals, it was too much information to use so I smoothed it out using mean and moving average calculations over days, weeks, days of the week and months.
From here, I created an algorithm to calculate who many people I would need on each day. I had to apply some managerial discretion here because the basic model would have had people standing around on the chance of something going wrong and this wouldn't be sustainable. After a couple of iterations, I settled on an acceptable ratio of pain to people.
With a greater understanding of the flow of passengers that came from the model, it also became apparent that the team needed slightly more depth in its structure. With the data available from the "pain-o-meter", I was able to show that two new positions could be created within our steady-state requirements and that these positions would support the supervision of the overall team when the airport was busy.
The Operations team was a combination of steady local people supported by a large, traditionally itinerant, group of working tourists. The model showed that we could expand our steady workforce within the new resourcing requirements.
In addition to a small cost saving, this change also allowed me to employ some of our more senior and talented temporary workers. Their addition to the team shifted the culture with an injection of enthusiasm and spirit. They also became our lead on-the-job trainers having been through their training in the previous year and having a great deal of empathy with our itinerant workforce.
Rostering Made Slightly Less Hard
The "pain-o-meter" was the base for rostering which could, theoretically, be completed six months ahead. I say theoretically because I still have a predominately itinerant workforce which couldn't be rostered effectively beyond two months.
But even this timeframe allowed the whole team to enjoy flexibility in their work schedule and comfort in scheduling leave in the future. While the model gave me a way of planning and rostering, it also gave the team some confidence in the roster and the rest of the team.
* albeit within capacity and slot constraints
** not really