Author: Chandrasekhar Jayaramakrishnan - Pre-Sales Consultant

Socrates once said that an unexamined life isn’t worth living. I’d go on to argue that an unexamined decision isn’t worth taking, especially if decisions are being taken based on an unexamined data set.

Statistical analysis of data can be powerful, but in Aviation it is often complex because of untenable assumptions made around the data. Not because of the lack of it, but instead due to its overwhelming abundance.

After having spent 8 years solutioning and implementing MRO systems, a move to Aircraft Data Analysis and consulting was a natural career progression choice for me. I was fortunate enough to have had a chance to work together with Sander de Bree, CEO of EXSYN Aviation Solutions beforehand as a 3rd party consultant in a project assisting the consolidation and Engineering transformation of a large helicopter holding company in Europe. I’ve followed EXSYN as a company since, and during a casual conversation with Sander around expanding EXSYN’s presence in the Asia-Pacific, and my interest in focusing on Aircraft Leasing solutions, we found the basis for an opportunity for me to join the growing team at EXSYN.

At EXSYN I particularly enjoy observing the Aircraft Data Consultants bridging technology with mathematics to extract cost-saving insights out of what is otherwise an industry filled with unstructured data. It takes a rare blend of consultants with experience in both Aviation and Applied Mathematics to do the same.  They work towards creating an organic structure for exchange and transfer of information between Airlines, MROs and Heli-Operators on data being managed across multiple systems. This varies based on the nature of the Maintenance Information Systems in place: MRO software, standard ERP packages, electronic data (.xls, .pdf, .csv) and unsurprisingly, paper.

Data Quality

It is great to see the determination of the EXSYN team in supporting their clients to improve their data quality. Throughout my career I observed that not enough attention has been paid and given to data audits that verify the information residing within either isolated data islands or integrated system in others: a process that intertwined with very many sets of inconsistent and redundant data in environments where multiple systems are in use. For example, while an MRO software could be in place to manage Base / Line / Shop Maintenance activities of an Airline / MRO, the Supply Chain support could be extended via an ERP, Flight Operations through a different package, with Human Resources and Financial functions residing in another arm of the IT food chain. Asynchronistic aside, the nature of information residing in these systems, even if pointing to the same “field”, are often very different.

While ATA Standards (Spec2000, Spec2500) have benchmarked approaches to make this process easier, we have found through our experience that even before Airlines and MROs think about adopting these standards, there is the perennial question around the clarity of data in use. Our blog post on “What to Expect from a Data Migration Project” highlights some of these issues, but the focus on this blog extends to how this information is being used by organizations to extract meaningful insights. And why the approach is futile if the data isn’t cleansed, enriched and validated beforehand.

Data Analytics  

EXSYN’s recent webinar on its Avilytics suite of solutions garnered an overwhelming response from the Aviation community. While it was encouraging to see their intent in going down the path of using Avilytics to churn out insights around Aircraft Reliability, Predictive Maintenance, Optimal Replacement time of components, Supply Chain trends among others, the insights are only as good as the source data that feeds in to Avilytics. 

More often than not, there are pointers and indicators that send our Aircraft Data Consultants alarming signals about the data not being “fit” enough for analysis. EXSYN has stressed on the importance of data quality in every engagement we take up with customers. For example, mis-match between ATA Chapters & Defect Descriptions render any Reliability Analysis run on that sample set futile. Incorrect part numbers, mostly a derivative of the actual part number with special characters, lead to redundancy when running Inventory Analysis. Configuration tree structures with substantial holes are of little use if the organization is using the source data for tracking component removals during heavy checks or running reliability analysis against the fleet.

The very nature of Avilytics is to use the data to predict the Reliability Trend or an AOG risk against a component(s) within a given time period. Analysis done with data not cleaned, audited and verified will yield incorrect results, thus it’s important to start with this first.

Practical Use Case: Calculating the Maintenance Cost against a Component over a given time

For example, if Avilytics is being used to predict maintenance costs against a Component over a given period of time, the cost would consist of four parameters considered for calculating the same: cost of inspection, repair, replacement and risks.  

The Cost of Inspection is usually a standard figure requiring minimal analysis. The Cost of Repair, on the other hand, involves the sum of basic repair cost (standard) and variable cost (based on availability / unavailability of spares, manpower, tooling). Calculating the variable cost here, as you can see, requires the data around the Inventory & Supply Chain to be “clean”: that is, an organic sequence of data linking the parts and the serial numbers, their current inventory value, history of purchase costs (including Purchase, Repair, Loan, AOG Orders), and lead time for procurement. Calculating the variable cost accurately feeds into the data being churned out by Avilytics around the Supply Chain metrics mentioned above: cleaner the data, more accurate the results.   

If the part has failed, and is BER (Beyond Economic Repair), the replacement cost is typically taken off the Parts Catalogue (and history of purchases against a preferred vendor).

The cost behind the risk, an important factor, is a derivative of the probability of the part failing within a given time window. Avilytics calculates the probability of failure of a component to fail within a given period in time, or a range (e.g. 70% failure between 2500 FH and 3000 FH). Once again, if the history data isn’t in place and in “clean” shape, this parameter renders an incorrect value.

Overall the common theme is that data needs to be clean, verified and enriched before either migrating it to a new system, or deploying tools such as Avilytics on top of it for analysis. There is a lot of work for us ahead and I’m looking forward to contributing to this with my knowledge and experience.

Next to that I’m keen to contribute to increasing EXSYN’s presence in the Asia-Pacific and Far East regions and to the growth in the Aircraft Leasing sector by structuring the existing solutions and consulting services to aid records consolidation during redeliveries and maintain compliance with contractual redelivery conditions. 

With Chandra joining the EXSYN-Team as Pre-Sales Consultant APAC we are taking the next big steps; as we are expanding our presence in the Asia-Pacific region and also broaden our business towards the aircraft leasing sector,  Sander de Bree, CEO, EXSYN Aviation Solutions