Digitisation and digitalisation are two confusingly similar but separate terms which are not interchangeable. It is useful to know the difference: digitisation is the conversion of analogue signals into usable digital information (see box) – a purely technical procedure. It also encompasses the scanning of documents and images into electronic formats, such as PDF.
Digitalisation is a much broader term, describing the process of moving away from using analogue data to inform processes; for example, rather than taking regular manual readings from a pressure gauge (analogue), using a pressure sensor to send digital readings to a data capture device (digital). This way, processes can be automated or controlled remotely, and long-term trends can be observed more easily.
Digitalisation can also refer to other aspects of the business, such as ensuring that documents are available on a cloud server. Business services provider Asite says: “Whereas digitisation is the process of making existing data and processes digital, digitalisation embraces the ability of digital technology to capture and assess data to make better business decisions and enable new business models”.
And this leads on to another widespread phrase: ‘digital transformation’. This embraces digitisation and digitalisation, and suggests a complete transformation of a business, using advanced tools such as machine learning.
OVERWHELMED
It’s important to establish how much information you need, and not end up swamped by data. Chris Hansford, managing director of Hansford Sensors, says: “Ideally you’d have everything monitored and uploaded to the cloud, but in reality there’s a cost implication.”
A lot of the time, says Hansford, “you’re trying to trend something over time to understand about a failure.” Sometimes, he says, “You’re not looking for a 1-2% change in vibration levels – you’re looking for a significant change. It’s about understanding the criticality of the plant and how it could fail.
“Some industries like the paper industry have been into this for a long time, and they will want to do a full analysis, so they take an AC sensor and put it into an online cloud-based system, and use FFT analysis to break down the frequencies and understand what’s going on,” Hansford adds, referring to Fast Fourier Transform, a useful mathematical technique (see also www.is.gd/usozox).
On the other hand, he says, “Something like a road tunnel fan has vibration sensors, but is it economic to have full vibration analysis? No, so they would have a 4-20mA output [connected to a control system]. This can send a warning to an engineer that a fan is going out of balance and they should clean the blades.”
Analogue signals from sensors tend to be to the 4-20mA or 0-10V DC standards, which are suitable for threshold on-off signals and some data readings. Programmable logic controllers (PLCs) typically have analogue inputs with ADCs built in, with a resolution of 12 bits. This outputs 212 or 4,096 possible values, so that a level meter covering a range of four meters would have a theoretical resolution of around 1mm.
However, analogue sensors such as strain gauges and thermocouples operate at very low voltages (sometimes 10mV or less) so they require specialised input modules which can perform appropriate signal conditioning, and which operate in ‘differential’ mode to compensate for electromagnetic interference (EMI or noise).
And PLCs do not offer a high sample rates: typically they can sample one or two values per millisecond, and this may be divided across inputs – fine for datalogging, but not suitable for applications like bearing vibration sensors, where high frequencies (>15kHz) may need to be measured.
Some modules can take analogue outputs and translate them directly into a digital signal following an industry-standard protocol. For example, Erbessd Instruments’ Phantom 4-20 GPIO can convert up to four 4-20mA sensor inputs at 16-bit resolution and output them in Modbus TCP/IP format via Bluetooth link.
Other digital data transfer protocols used in industry include Profinet and Ethercat, both of which use the same hardware as Ethernet, albeit usually with ruggedised cables and connectors.
Devices known as ‘gateways’ can translate between networking standards, allowing different manufacturers’ devices to talk to each other.
However, the trend is towards fully-digital sensors, with conversion on board and an output direct to a network, wired or wirelessly.
BOX: DIGITISATION BASICS
The usual method of converting a varying analogue signal to digital is pulse code modulation (PCM): the incoming signal (usually a voltage from a transducer or sensor of some sort) is periodically measured (sampled) and converted into binary.
PCM is used for most kinds of digital audio transmission and recording – in this case, the microphone is a transducer to convert varying air pressure (sound waves) into voltage. The small voltage from the microphone may be pre-amplified before it goes to an analogue-to-digital converter (ADC or A2D) which slices the signal into ‘samples’ covering a particular time period, and measures each in turn.
The two critical criteria in PCM are the sample rate and the bit depth (or resolution). Sample rate is directly linked to the frequency range which can be measured by the PCM system: the maximum signal frequency (in Hz) is half the sample rate (also Hz, or samples per second).
Bit depth (or resolution) refers to the number of binary digits used to measure each sample. Each extra bit doubles the possible number of values: a four-bit system can measure 24 or 16 possible values, from 0 (0000 in binary) to 15 (1111). A CD’s 16-bit system has 216 (or 65,536) possible values. Using more bits gives you more resolution – more precision – but it also adds to the processing and data handling load.
The incoming analogue signal should be matched to the ADC so that the maximum possible signal (typically voltage or current) gives a digital value equivalent to the bit depth; this is one element of signal conditioning. So if you want to measure signals from 0-20mV and your ADC accepts 0-10V, you need a voltage gain of 500 (often expressed as a ratio in decibels, in this case 54dB). For signals that are too high, the ADC can only return its highest value – effectively giving a false reading known as clipping. To prevent this, you can limit the peak analogue signal (still a false reading, but unlikely to damage the ADC) or reduce the gain. On the other hand, insufficient gain means you are not making use of the full bit depth, losing resolution. And low-voltage analogue signals are vulnerable to electronic interference, requiring care with wiring, earthing and shielding.
The good news is that all of these functions can now be integrated into a single tiny chip, and built into almost any sensor to output whatever signal is required.
BOX: DIGITALISATION AND SUSTAINABILITY
An international survey of 765 business and technology leaders commissioned by ABB, available via www.is.gd/qulawu, revealed that while 96% believe digitalisation is “essential to sustainability,” just 35% have implemented industrial IoT solutions at scale.
This gap shows that while many of today’s industrial leaders recognise the important relationship between digitalisation and sustainability, the adoption of relevant digital solutions to enable better decisions and achieve sustainability goals needs to accelerate in sectors like manufacturing, energy, buildings and transport, according to ABB. Perceived cybersecurity vulnerabilities were found to be the number one barrier to improving sustainability through the Industrial IoT.
Peter Terwiesch, president of ABB’s process automation business area, said: “Sustainability goals more and more are a crucial driver of business value and company reputation, and Industrial IoT solutions are playing an increasingly important role in helping enterprises achieve safe, smart and sustainable operations.
“Unlocking insights hidden in operational data holds the key to enabling literally billions of better decisions throughout industry and acting upon them, with significant gains in productivity, reduced energy consumption and lower environmental impact.”
ABB reports that more than 70% of its R&D resources are dedicated to digital and software innovations. It claims to offer more than 170 Industrial IoT solutions. Its ABB Ability portfolio of digital solutions is said to enable a host of industrial use cases to make improvements in energy efficiency, including condition monitoring, asset health and predictive maintenance.
DCS AND SCADA
Two useful and related concepts are DCS and SCADA: that is, distributed control system (DCS) and supervisory control and data acquisition (SCADA).
Both terms describe process monitoring and control systems which involve multiple sensors, controllers and actuators, as well as HMI (human-machine interface) devices which allow operators to communicate with the system.
The point of DCS is that these devices are connected to local processing nodes which are distributed around the system, rather than to a single central processor or control unit; this should make the system more robust overall (if one processor fails, the rest of the system can carry on) and it reduces any effects of lag due to distance from the controller (although this is less of an issue as wired and wireless data links become faster and more reliable).
SCADA systems tend to have a centralised processor, and to be more event-driven – that is, the system reacts to events rather than continuously adjusting the process. It often also includes a software or hardware ‘historian’ which logs all the data aggregated by the system to produce analysis and reports.