Helicopter operators are increasingly data rich, but how much of it is producing value to safety, efficiency and effectiveness?

Aeroplane manufactures and airlines have been using their data to improve efficiency, reduce costs and have attempted to improve the passenger experience for a while. How can the major helicopter operators catch-up and what are the benefits?

Whilst the offshore helicopter industry has pioneered the use of Health and Usage Monitoring System (HUMS), other areas have not advanced as far. Too often we see that data is held in silos without being readily accessible to other systems. So what are these system silos? Apart from HUMS, operators will have a safety reporting system, engineering and supply chain system such as SAP, aircraft Tech Log data, Flight Data Monitoring, flight planning tools, crewing and rostering records.

Measuring drives behaviour as it indicates where problems exist.

What operators lack is a mechanism to find the Rumsfeldian “unknown unknowns” and the unusual patterns or outliers in their data. Flight Data Monitoring (FDM) is an example of a data rich system which is mainly used for compliance and yet it has the capability to do so much more. As well as having fixed thresholds to identify errors in the flight profile, the data can be used to identify “Normality” and anything outside those parameters might be an issue. For example, are the outliers evenly distributed across the operations or are they more prevalent at certain locations, at certain times of day, on certain aircraft or with certain crews? Using this methodology, issues with approaches to problematic locations, fatigue issues, with certain aircraft types or with training can be identified for further investigation by specialists. An example of using FDM data to identify normality is an ongoing UK CAA project to define enhanced HTAWS warning envelopes for offshore operations[1] .

Using FDM as a compliance tool tends to treat each event as a singular event with little widespread learning. Further analysis of the FDM data can be used to map turbulence around a helideck. Mapping the turbulence environment will provide a feedback loop to the Helideck Landing Limitations and CFD/wind tunnel modelling done when the installation or ship was commissioned. The military spend large amounts of money on Ship Helicopter Operating Limitations (SHOL) trials, which are by necessity a snap-shot of conditions over the period of the trial. Using FDM data provides the equivalent of an ongoing SHOL which captures all conditions over the life of the vessel with minimal additional cost.

FDM can be used to record systems faults or inappropriate combinations of autopilot modes. One operator recorded autopilot faults and found that crews had only reported around 10% of the dropouts, indicating that they had normalised the issue. By comparing recorded faults to the safety reports submitted about those issues gives a good indication of the reporting culture in that organisation. The reporting rate versus events captured in FDM should be a KPI for every operator. Measuring the way that autopilot modes are used can indicate the compliance rate with SOPs, which has been a factor in both aeroplane and helicopter automation related accidents.

By additional analysis of the FDM data the actual risks the operations encounter can be quantified, providing a feedback loop to the organisation’s Safety Management System (SMS). In the example of the autopilot faults above, the actual failure rate was many times higher than that assumed in the Predictive Risk Assessment and so the true risk to the organisation was much higher than initially assessed and needed action.

Operational data provides a means to optimise operations. How many operators measure their operations against KPIs such as seat cost per mile? With the offshore helicopter market under such tight cost pressures are they sure that their aircraft are being operated in an efficient and effective way? By optimising the departure times of the flights, cruise altitude and airspeed savings can be made. Some fleets such as the S92 have a large variation in their Declared Operating Mass of greater than a 1,000 lbs between old aircraft and the newer heavier versions. Are the lighter aircraft being used on the longer flights where they can maximise the client payload?

Tech Log and Supply Chain data can be used for component reliability modelling. Most operators leave it to the OEMs, but smart customers will be able to resolve issues quicker if they identify problems at an early stage rather than sitting back and waiting for OEM action. Part M requires organisations to “assign responsibility for co-ordinating action on airworthiness occurrences and for initiating any necessary further investigation and follow-up activity to a suitably qualified person with clearly defined authority and status”, but how many do it effectively with no room for improvement?

Safety Reports can provide a wealth of information, but how well are they analysed? Tools such as Cluster Analysis can draw out areas for further investigation by specialists.

Of course there is a worry that Big Data = Big Cost, but this need not be the case.  The additional analyses discussed above are not mandated and are about gaining extra insight into the company data, it does not have to run in real-time or be safety critical. Start small and gradually build up a capability. After all simple Machine Learning and Text Mining can be done in Excel, Open Source tools in languages like Python can also be extremely powerful and yet  low cost, running on a stand-alone laptop with analyses being run every few weeks; after all the hidden trends are not likely to change quickly. Virtually all standard systems can export data as flat files in formats such as csv which Open Source tools exploit.

However, any company wanting to start this process will have to work with their IT Department, where the first barrier to innovation is likely to be met. Large IT Departments love “Enterprise Systems” which allegedly reduce risk to the business, whilst in reality they initially reduce risk to the IT Department as they can buy in a system and contractors to do the work. Start the analysis as a trial using free tools and when the benefits are seen, by all means upgrade to a full Business Intelligence system, but it is not necessary to spend a lot at the outset.
 
Of course smarter use of data is not a silver bullet. However, improved insight of what data is telling a company should help the specialist improve their areas, where complex patterns are often missed due to workload or insufficient visibility of the underlying trends. British Cycling has dominated the last few Olympic Games by the “aggregation of marginal gains”. If everything you do is improved by 1% then those small gains add up to remarkable improvement (and lots of gold medals). So try breaking down those data silos and aggregate the gains that will come from improved information and insight.


[1] http://publicapps.caa.co.uk/modalapplication.aspx catid=1&pagetype=65&appid=11&mode=detail&id=7887