Rich data to make everyone and everything smarter

We have all produced so much information in recent years that we’ve entered the zettabyte era. By 2018, every storage specialist will need to maintain three times as much data as today. Do companies really need to sit on that much information? We see three reasons to answer favourably: one – some are obliged to in order to comply with sector or country regulations, two – the price per terabyte is plummeting and three – information can be transformed into insight and insight into value. Rich data leads to knowledge and insights.

According to Bain and Company, companies with advanced analytics make better decisions, take them faster and execute them better, leading to better results and value. So if we want to achieve business value, we need to convert raw data. Structure the unstructured. Classify your data, run the analytics and make everyone and everything smart..

From corrective maintenance to predicting failures four hours in advance

Running analytics on both old and current data allows you to predict the evolution of machinery, and turn corrective into prescriptive maintenance. Preventive maintenance is among other things based on condition-based monitoring information (pressure, oil, etcetera) from the machine itself. It often requires important technical interventions and thus costs more money than absolutely necessary.

With big data you can organize maintenance based upon other indicators and plan ahead, which is why BASF did a feasibility study on predictive/prescriptive analytics. “We wanted to optimize maintenance and the functional performance curves of the equipment. Those decrease over time”, said Jürgen Moors, Head of Technical Verbund Management at BASF.

“Data scientists make old pumps speak.”

His team took a high pressure, 30-year old liquid ammonia pump as an example as proof of concept. Process data like pressure and temperature data had been stored for years and years but left untouched. Moors’ team added the maintenance plans and quality data from BASF’s ERP software, and fused in business knowledge.

Indicators for five common causes for failures

The team analysed the past failures and found out there were five common causes. BASF now knows what to change when one of these five fingerprint causes occur. “In 80% of the cases, we could predict the failures four hours in advance”, Moors explained. “That’s important, because you need the time to slow down the process. This could avoid clogging of solids in the piping for processes where solids are involved. We know when to lower the loads and are able to lengthen the life span of machinery. Also, with targeted repairs, you don’t look at everything anymore. Maintenance can be quicker and more cost-efficient.”


BASF’s 3 take-aways for doubters

1. There’s always data somewhere: you can start today . Make your organization data-conscientious and data-aware – you can already achieve a lot with qualitative data
2. There are huge opportunities in every sector : energy, insurance, consumer and now also manufacturing.
3. If you embark in this, technology is an intermediate step. You need preparation, group the required skills for the entire plant, both cross-functional and cross-department, and build new processes using analytics, e.g. maintenance