Deriving real value from industrial IoT projects

Deriving real value from industrial IoT projects

July 2018

By David Griffin, Senior Consultant at 42T

We all recognise hype when we see it. Every trade journal seems compelled to feature an article linking their specialist area with the latest big thing – be it machine-learning, graphene or blockchain.

Deriving real value from industrial IoT projects

One of the by-products of this effect is that companies can feel they are in danger of being left behind if they are not doing something relating to new technology.  And that can lead to initiatives that may be created for the wrong reasons.

Industrial IoT is in many ways an example of this.  A number of companies have seen edicts passed down from board level that products must be connected in xx years time.

This is never an ideal starting point for any project:

  • a project that arises without a clear strategy behind it can be cancelled just as easily
  • any project without a clear tangible benefit in mind will lack the context within which to sensibly make the inevitable technical or commercial decisions.

This is a pity because there are many tangible benefits to be had from connecting products, processes and assets to gather more granular information about customers, costs and organisational performance.

Three timelines for payback

Benefits can often be broadly categorised into 3 levels of RoI

1.   Immediate and certain payback

2.   Delayed but fairly certain payback

3.   Long term and somewhat speculative payback

The first category might include, for example, a situation whereby a person used to manually record data (perhaps for regulatory purposes, such as monitoring hot tap water temperatures for legionella compliance in a large building), but as a result of the IoT installation the data is now gathered by installed connected sensors automatically.  The moment the monitoring system goes live, the labour costs fall.  Immediate payback.

The second category might involve monitoring a process to spot when something has stopped functioning correctly.  This could result in an immediate alarm which allows the problem to be addressed quickly rather than when expensive knock-on problems have piled up.  Though it may not save any money on day 1, if the original justification was based on solid analysis of past events it is only a matter of time before the system starts to earn its keep.  Though delayed, the benefits will be seen.

The third category of project is one where the ultimate benefit is something predictive based on analysis of large amounts of data (or worse, where there are anticipated benefits that are not substantiated by robust analysis).  For example, “by monitoring motor current on all the conveyors, we’ll learn how to predict motor failures and so we’ll be able to change out motors before they fail”.

The challenge

With these projects, the challenge is that a big part of the learning required to deliver on the promise only starts once all of the system has been delivered.  And what is more, there’s no cast iron guarantee that the hoped for benefits will all be delivered.  In other words, the connectivity is a pre-requisite for the further project, but does not alone deliver a payback.

If the IoT system envisaged is a product or service being sold to other organisations, there’s a further unknown in the risk mix.  Even if the promised results can be delivered, will paying customers choose to invest?

Projects in this third category may actually give the best long-term payback and create the most competitive advantage but are harder to move through a typical corporate project justification process.

In an ideal world, an investment in connected sensing will give payback across the range of categories described above.  The initial investment can be justified against some tangible immediate benefit that doesn’t require 100% coverage to start paying back. And the longer term higher level payback builds as the reach of the connectivity spreads.  If this is not the case in a proposed project, it may even be worth changing the structure of the project slightly to enable some short-term payback along the route to the final vision.

Don’t be limited by what’s already there

Often a project to add connectivity in industrial processes is based on connecting process equipment that already monitors the requisite parameters.  This pre-supposes that:

a)   the parameter of interest is already being measured and

b)   the equipment control system already lends itself to connectivity.

In some cases, it can make more sense installing dedicated separate connected sensors that don’t connect to the existing control system in anyway.

This might be because:

  • the existing control system is proprietary and not accessible
  • the existing control system is validated and cannot be modified without revalidation
  • the set of required parameters spans multiple machines and/or includes quantities not actually measured by any existing processes. Eg. the ambient temperature around an open belt where product cools between processes – not officially a control parameter but potentially relevant to final product quality
  • some or all of the equipment is either not networked or else not even under computer control.

Creating a connected monitoring system using extra sensors makes it possible to create a centralised dashboard of the key factors which may affect process performance without any disruption to the existing line’s output.  Such an approach can offer many benefits:

  • early warning when someone inadvertently makes a key parameter change or the properties of line infeed change significantly, allowing offsite engineers to rapidly respond before a large amount of output is wasted, even when the process is not computer controlled
  • the opportunity to learn whether parameters related to the environment around the actual process are correlated with fluctuations in output
  • the means to understand stabilisation time after line start-up which may not be fully represented by the machinery’s internal measured parameters

Gather the right data

The key to achieving these benefits are:

  • understanding which parameters affecting the process are most important to monitor (bearing in mind that some of them may not actually be intended process parameters)
  • determining the most appropriate way to measure the data
  • cost effectively, reliably and securely transferring the data in a usable form to a centralised repository from where it can be analysed and visualised.

Understanding which parameters affect the process will involve combining knowledge of the intended process behaviour with a study of the broader context in which the process operates.  This might include the environment outside the process, the behaviour of upstream processes, and the variability of raw materials used in the process.  Anecdotal evidence from experienced operators often gives ideas of where to look for clues.

So what characterises a successful project?

In summary, a successful IIoT project might be characterised as one that:

  • Can demonstrate at least some tangible payback as soon as the first implementation is complete
  • Prevents wasteful events or activities by putting more timely, more complete or more accurate information in the right people’s hands as soon as it becomes available
  • Makes local data available globally, or makes new insight possible by allowing data to be combined together that were previously not known, or else not all in the same place at the same time, such that engineering data can be correlated with commercial performance.

If one of more of these can be convincingly demonstrated in advance, the future is bright.

Share this article:

Background

What will you ask us today?

We believe in asking the right questions to drive innovation; when we know the right questions, we generate the ideas to answer them.