29 May Accuracy & Precision: 2° PART
In this second and final part, we would like to introduce the concept of Accuracy, how to “read” these attributes and some final considerations about DOTECO products.
II° PART: ACCURACY
Where are you aiming?
After the last time, you decided to not give up and kept training to improve your precision.
Today you will test your archery skills in your first tournament by shooting 50 arrows to a target. Good luck!
At the end of the session, you review the performance with your trainer.
Here’s how the target looks like:
On one hand, all the arrows ended almost in the same spot, however none of them hit the center of the target.
So, how would your performance be rated?
The response is: high precision but…. very low accuracy!
Accuracy is, on average, how far is the real output, or result of a process (in this case the arrows shot) from a pre-set value (you were expected to hit the center of the target); even if you managed to send all the arrows close to each other (high precision), all of them landed way off the desired area (low accuracy).
Accuracy and Blenders
As we mentioned in the previous article about precision, there are many similarities between archery and dosing processes. A good performance doesn’t only involves that the arrows lands as closer as possible, but mainly it is required that they hit the center of the target.
In the same way, a good blender must be able to dose always the same amount over time (high precision) and in the desired quantity (high accuracy).
Let’s see an example:
In order to be able to monitor the accuracy, first it is necessary to set the recipe and choose a desired throughput (kg/h), then at constant intervals of time we take samples of one of the ingredients and weight them.
Once we have a sufficient amount of samples, let’s visualize them by displaying in a chart as shown below: the blue dots represent the single measurements (samples), the red line indicates the pre-set value and the green one the average of the observations.
It is immediately noticeable that, despite the samples are all very close to the average (good precision), the whole data collected is shifted from the set-point. Same as in the archery example next to the chart, the arrows landed close to each others, but very far from the center of the target.
In other words, the blender is dispensing the ingredients always in the same amounts and proportions, but not the ones we wanted!
Let’s try again, but this time we will use an high end blender.
Compared to the first test, the samples average (green line) are now very close to the pre-set value (red line)*.
Now the blender is dosing the expected amount and percentage of ingredients.
*the distance between the red and green line has been exagerated for visualization purposes.
Here is another simple numerical example to show the effects of accuracy on the ingredients percentage.
On the first column are listed the recipe ingredients, on the second one the set point percentage – in other words the desired proportions of each ingredient – and in the third one, highlighted in red the actual percentages for each component. The higher will be the difference between the actual value and the pre-set one the lower will be the accuracy, the higher will be the cost of dosing not necessary additives.
How to calculate accuracy
Thanks to modern technology, blenders’ accuracy and sigma are automatically calculated by computers.
However, now that we understand what that “number” represent, might be interesting to learn also where it comes from.
Accuracy, is the ratio between the average set-point (the measurements average) and the set-point (the desired output).
The lower the value will be, the higher the accuracy.
Low accuracy, high costs
Accuracy might have a relevant impact on the final outcome.
A low positive accuracy, results into an excess of material dispensed, which will translates in an higher and not needed extra material consumption.
On the other hand, a low negative accuracy means that the ingredients will not be dosed as required and the final product will not meet the specifications.
Accuracy and Precision
The picture below combines precision ad accuracy at different levels. The horizontal axis represent the accuracy level from low to high, while the vertical one depicts different degrees of precision.
All that glitters is not gold
Now that we have a better idea of what precision and accuracy mean in dosing processes, how can we correctly interpret performance data?
There are some conditions to be taken in consideration to be able to correctly judge blenders quality performances.
Technological and natural limits: technological development reached very high levels, hovewer it is still a bit far from absolute perfection. Considering for example load cells resolution, vibrations and general tolerances.
Numbers should be contestualized: sure it is of great impact to display low percentages on a brochure, but without a context, in reality, there’s little to no meaning. Precision and accuracy vary, depending on various factors such as the material type, environmental conditions and throughput rate.
For example, the smaller the throughput rate the higher will be the impact of a variation in the dosing weight.
Class of accuracy: while reading specifications sheets, do not confuse blenders accuracy and precision level with the so called “load cells class of accuracy” (the error tolerance of the load sensors). In reality accuracy and precision are the result of many different factors: how it is mechanically engineered, the quality of the electronic components and the software architecture.
Homogeneity: There is a 3rd key factor that greatly influence the output quality, the ability of the blender to mix all the ingredients homogeneously. A well designed mixer, can contribute to lower the incidence of dosing errors, consequently helping to increase the overal blender performances.
With the ADROIT series, Doteco has reached new heights in dosing performances.
The software: the core of the blender, now includes specific and complex predictive algorithms to reduce fluctuations and increase the output consistency.
GRADO ADROIT is equipped with:
- High resolution load sensing cells (24 bit – 0,004 gr), installed on vibrations and shocks absorbers.
- Low inertia technopolymer slide gates, allowing the open/close transition much quicker compared to other systems.
- Specifically designed mixer to grant homogenenus batch blending and to avoid material stratification.
- High resolution load sensing cells (24 bit – 0,004 gr).
- High performance dosing screws: low tolerances, reduced friction, inclined to avoid accidental material drops.
- Cascade Mixer: special design and gravity, jointly provide an incomparable homogeneity.