In the previous post (Blog post 5), we looked at direct sensor data acquisition. In this blog post we will examine another type of data acquisition : Derived measurements
What is a derived measurement ?
Derived measurement is an indirect method of data acquisition that is used when direct measurements are not possible. Derived measurements are typically done where direct measurements are not feasible due to
Physical constraints - Physically not possible to install a sensor or any other means of measurement (or) installation requires considerable modifications to the equipment
Technological constraints - Right type of sensors not available
Cost constraints - Direct measurement is too expensive
Derived measurements are typically done in two steps.
Step 1 entails identifying and measuring a proxy parameter that has reasonable correlation with the target parameter that needs to be measured
Step 2 involves calculations to be done at the edge layer to derive the target parameter
To better understand derived measurements, let us explore further with a simple example.
Application Requirement : Deriving drilling count from drilling machines
Background : In a machine shop, Industrial drilling machines are used for drilling thick steel plates. These drilling machines are driven by electric motors and are in operation for the past 20 years. These drilling machines are manually operated. There are no sensors available in these machines that can provide usable data. Considering the age of these machines, it is extremely difficult to install required sensors to directly measure the drilling count
Objective : Measure the drilling count (i.e.) no of drills performed by the drilling machines. This data is required to calculate and enhance the OEE[1] of drilling machines
Here, we have a requirement to measure the number of drilling operations (target parameter) without the possibility of direct measurement. So, we have to perform a derived measurement.
Step 1: Identifying and measuring the Proxy parameter
Considering the fact that the drilling machines are very old, manually operated, and motor driven, electric current consumption is one of the more suitable non-invasive proxy parameter for deriving the drilling count. Now, how do we measure the current consumption? Shown below (Fig 1) is one of the possible solutions.
Fig 1 : Measuring electric current consumption of the drilling machine
Different components that are required to do this derived measurement are shown in Fig 1.
Three current transformers (CTs) are affixed to the cables that supply power to the drilling machine.
These current transformers proportionally step down the AC current consumed by the drilling machine to a standard amperage value.
This stepped down current along with the supply voltage are given as inputs to the Energy meter.
Energy meter measures and calculates electric current consumption and all energy related parameters and provides an the data to the IoT gateway (Edge device) through Modbus protocol.
Now that we have looked at a possible method of measuring the current consumption of the drilling machine, let us figure out a way to derive drilling count from the electric current consumption.
Step 2: Calculations at the edge
Shown below (Fig 2) is a simplified version of the electric current consumption signature of the drilling machine while in operation. Y-Axis is the average current consumption of all the three phases in amps. X-Axis is time in seconds.
The peaks denote drilling operations, and the troughs denote idling of the drilling machine. Other things being constant, the height and duration of the peak depends on the thickness of the steel plate that is being drilled. The thicker the steel plate, the taller and wider the peaks.
Fig 2 : Drilling machine current signature
So, detecting a drilling operation is a function of detecting a step change (positive edge) in the electric current consumption. Please note that since the change in electric current consumption will be in the range of milliseconds it is imperative that this positive edge detection is done real time.
Detecting and accumulating the drilling operations to derive the cumulative drilling count can be done at the edge layer (IoT gateway) with a simple application program.
Derived measurements : Key considerations
What we saw with the drilling machine is a straight forward example of derived measurements. Please note that there are no standard methods for doing derived measurements. Choosing and measuring the proxy parameter and the edge computation will vary from case to case. It will be helpful to consider the following points before going ahead with derived measurements.
Examine the need for derived measurement – Is direct measurement not possible? Why so?
Analyze the suitability of derived measurement – Derived measurements are almost always not as accurate as direct measurements. Proxy parameters can be influenced by other variables. For instance, in our drilling machine example the current values can be influenced by sudden voltage fluctuations. In-spite of these factors, will derived measurements still suit your application needs?
Compute at the Edge – Most of the times, derived measurements need an intelligent edge device for computation. Evaluate the feasibility of having compute at the edge before going ahead with derived measurements
Functional knowledge – Derived measurements need a good understanding of the functioning of the machine / equipment / process to enable choosing a proper proxy measurement parameter and the right computation.
In blog posts 5 and 6, we have explored two types of data acquisition; Direct sensor data acquisition and Derived measurements.
Check out my next blog post on acquiring data from Industrial control systems like PLCs and SCADA systems ...
[1] OEE or Overall Equipment Effectiveness is an index that is very widely used in the manufacturing industry. OEE is an indicator of manufacturing productivity. It is a multiple of Availability, Performance, and Quality. Higher the OEE, the better the effectiveness and productivity of the equipment. OEE of 85% and above is considered as very good.
Great blog sir, awaiting next post!