Professional Documents
Culture Documents
Department of Engineering,
University of Perugia,
Via G. Duranti 63,
Perugia 06125, Italy
e-mail: luca.landi@unipg.it
Alessandro Stecconi
Department of Engineering,
University of Perugia,
Via G. Duranti 63,
Perugia 06125, Italy
e-mail: alessandro.stecconi@unipg.it
1 Introduction impact resistance Y can be interpreted also like the so-called v50,
i.e., that impact velocity for which there is the same probability
In the design of safety guards, the designer aims to determine
that complete penetration occurs (i.e., the projectile do not com-
the impact resistance Y of the guard in relation to the thickness
pletely pass through the guard).
and material of the guard. However, the normative documentation
There are also specific type-C standards for certain types of
that defines the state of the art for the dimensioning of safety
machines, such as ISO 16090-1 [2] (2017) or ISO 19085-1 [3]
guards today presents different approaches in the definition of the
(2017), in which the validation conditions for impact tests are dif-
impact resistance Y.
ferent. In this type of standard, the impact resistance, again associ-
In general, the reference standard that defines the testing and
ated with a guard of given thickness and material, is that impact
validation methodologies of ballistic impact tests on machine
velocity for which a plastic buckling occurs without a passing
guards is the type-B2 standard ISO 14120 [1], Annex B. It defines
crack. This type of validation condition, which is certainly more
the main parameters that characterize an impact test, such as the
precautionary, is generally preferred in the industrial field to
shape and size of the projectile, the impact velocity, the size and
define the suitability of a guard to be mounted on the machine
thickness of the panels to be tested. The definition of impact
tool. For the sake of clarity, in the rest of the paper, we will define
resistance Y can instead be identified in the validation conditions
as impact resistance Y the condition defined by the speed v50 and
of the tests performed. In fact, according to Ref. [1], the test is
Ysf, as safe impact resistance, the speed for which no through-
failed if the projectile passes through the test object (e.g., material
crack occurs at all (even if the through the crack is very small the
sample, guard, etc.). In other words, according to Ref. [1], the
test is failed). In different standards for the safety of machinery,
both v50 and Ysf are generically referred to as impact resistance.
The common problem with the different standards listed above
Manuscript received May 28, 2021; final manuscript received November 2, 2021; is the number of tests required to determine the suitability or oth-
published online December 14, 2021. Assoc. Editor: Alba Sofi. erwise of the panels. In fact, it only takes one passed test to
ASCE-ASME Journal of Risk and Uncertainty in Engineering Systems, JUNE 2022, Vol. 8 / 021001-1
Part B: Mechanical Engineering C 2022 by ASME
Copyright V
determine that a panel of a given thickness and material is suitable The influence of these factors on the impact resistance Y of
for installation. machine guards has been investigated in other studies. In the study
It is now well known, thanks to the numerous studies and of Landi et al. [5], the influence of inclination is considered when
experimental impact tests carried out over the years, that this using the standardized projectile shape. The influence of inclina-
deterministic approach is too simplifying and limiting for an accu- tion is decisive, so that inclined impact tests must be rejected for a
rate assessment of the effective impact resistance of guards. In proper ballistic limit vbl assessment.
particular, one single shot is not sufficient to describe the overall As an example, Borvik et al. [6,7] and Landi et al. [8] per-
penetration resistance of a guard, because many physical parame- formed many experimental tests, respectively, on steel and poly-
ters can affect the guards and projectile behavior. The main crit- carbonate sheets of different thickness dpm.
icalities that characterize the current standardized tests will be Also, the influence of aged polycarbonate sheets has been tested
described in the following State of the Art section. using the bisection method for the determination of the impact
A method used in industry for greater accuracy in determining resistance Y, see Uhlmann et al. [9,10].
the impact resistance Y is the bisection method. According to this Finally, regarding the point of impact, which may not be coinci-
method, by performing multiple tests, the impact resistance Y of dent with the center of the target, the research conducted at IWF
guards is determined by delimiting the range between pass and by Bold [11] showed that the center of the polycarbonate sheet is
fail of the test model with a number of impact tests. However, for not the worst point for the penetration. In fact, the kinetic energy
a small number of impact tests the range may be too wide to deter- Ep required to perforate the guard decreases as the point of impact
mine the impact resistance Y with an exact value. In addition, approaches the boundary, see Landi et al. [4,5] at least for poly-
uncertainties in measurement and material properties are not con- carbonate, where the deflection capability of the plate is reduced
sidered. This approach is limited by a number of relevant con- near the borders.
straints that should be considered and analyzed in order to achieve The normative annexes introduced in type C safety standards,
a better design of machine guards with regard to safety. thanks to a more precautionary impact test validation method, led
Scientists at the University of Perugia in Italy in cooperation to a remarkable reduction in the number of accidents, for referen-
with the Institute of Machine Tools and Factory Management ces see Ref. [11]. Although the reduction of injuries and deaths
(IWF) of the Technische Universit€at Berlin in Germany investi- can be seen as a degree of the appropriateness of the safety stand-
gated a new approach to determine the impact resistance Y by a ards developed, there are factors that need to be considered in
probabilistic method using the Recht and Ipson (R&I) ballistic order to further reduce the number of accidents. The determina-
equation. This method considers multiple impacts to determine a tion of the impact resistances Y and Ysf with multiple impact tests
ballistic limit vbl, where the projectile is close to fully penetrating and the influence of the projectile head shape, material, and thick-
the test pattern and that can be considered equivalent to the impact ness dpm on the impact resistance Y can be retrieved in a statistical
resistance Y as previously defined. Therefore, indications of way. A degree of uncertainty about the accuracy of predicting real
uncertainties and identification of outliers during the impact tests behavior based on the deterministic approach of existing standar-
are possible. In dependence of the ballistic limit vbl, the safe dized tests and an alternative probabilistic approach is presented
impact resistance Ysf can be retrieved by defining a velocity reduc- in the following sections.
tion coefficient cR,v. The results are used to implement a more reli-
able method to determine the impact resistance Y and the safe
impact resistance Ysf for the standardized tests with an indication 3 Probabilistic Approach Based on Recht and Ipson
of uncertainties. In brief, the authors aim is to measure, using a Estimation
reliable and reproducible statistical approach, the uncertainty
between the two verification methods proposed by the standards. 3.1 Introduction of the Recht and Ipson Method. It follows
from the above considerations that a more detailed analysis of the
impact resistance Y is required. The resistance of a material to
penetration is often considered in terms of the ballistic limit vbl
2 State of the Art defined as The minimum velocity required by a projectile to com-
Impact tests can show variations in results of the same test setup. pletely penetrate a target, see Recht and Ipson [12] and Borvik
Landi et al. [4,5] investigated the variation of single-trial attempts. et al. [6] as references.
For the evaluation of uncertainties, the National Institute for Insur- The ballistic limit vbl can be defined deterministically as well as
ance against Accidents at Work, which deals with finding a rough probabilistically. The deterministic approach used in standardiza-
index for the repeatability of tests, investigated the uncertainties of tion, as described in the previous sections, leads to not entirely
impact tests where two consecutive standardized attempts, per- reliable results by defining a reference value for the ballistic limit
formed under the same conditions of initial velocity vpr,i, projectile vbl or the impact resistance Y.
shape, target material, and thickness dpm, led to different results. Var- Therefore, a statistical approach seems necessary to evaluate
ious attempts were carried out according to ISO 14120 [1], Annex the dispersion of results in determining the impact resistance Y by
B, using a projectile mass of mpr ¼ 0.1 kg, an initial velocity implementing a regression curve. The most commonly used equa-
vpr,i ¼ 70 m/s, and guard of low carbon steel with a thickness tion for determining the ballistic limit vbl is the R&I in the follow-
dpm ¼ 2 mm. The materials tested were steel sheets of S235 and ing equation:
DD11, which are widely used for the structural design of mechanical
components. Three consecutive tests were carried out for each of vpr;r ¼ aðvppr;i vpbl Þ1=p (1)
them. The DD11 passed 2 out of 3 tests (no penetration in two cases)
while the S235 passed 1 out of 3. These very simple test results con-
firmed the evident dispersion in terms of the impact resistance Y of where vbl is the ballistic limit, vpr,i and vpr,r are, respectively, the
the guards and the unreliability of the single-attempt test. initial projectile velocity before and the residual projectile veloc-
In general, standardized tests are not capable of assessing all ity after impact, when the target is fully penetrated; p and a are
circumstances that may affect the influence of a tool or working two dimensionless parameters, see Landi and Amici [13]. Specifi-
material on a machine guard. There are some relevant factors that cally, a is defined by the ratio mpr/(mpr þ mpl), where mpr is the
can significantly affect the results on the industrial environment: mass of the projectile and mpl is the mass of the plug ejected from
curved shape of the guard, the plate following impact. In general, the ballistic limit vbl is
nonperpendicular projectile impact to the target, evaluated by a method of least-squares approximation based on
eccentric projectile impact, and several impact tests with different initial velocities vpr,i. Figure 1
standardized projectile without rotational symmetry. shows an example of a typical R&I curve.
Fig. 1 Exemplary visualization of the R&I curve Fig. 2 Standardized projectile shape according to ISO 14120
ASCE-ASME Journal of Risk and Uncertainty in Engineering Systems, JUNE 2022, Vol. 8 / 021001-3
Part B: Mechanical Engineering
Table 1 Standardized projectile mass and dimensions accord-
ing to ISO 14120
0.1 20 10 10
2.5 50 30 30
PC sheet PC sheet
Parameter thickness dpm ¼ 10 mm thickness dpm ¼ 12 mm
a 01.00 00.99
p 01.70 01.87
Fig. 5 Experimental setup vbl (m/s) 58.50 64.20
Ubl (m/s) 59.55 65.57
Lbl (m/s) 57.47 62.31
ems 04.49 11.57
Fig. 6 Exemplary impact test results for PC: (a) passed and (b)
failed
ASCE-ASME Journal of Risk and Uncertainty in Engineering Systems, JUNE 2022, Vol. 8 / 021001-5
Part B: Mechanical Engineering
the best fit. It is important to be aware that test 7 in Table 2 and
test 8 in Table 3 delivered results according to the validation con-
ditions of standardization with only bulging (permanent deforma-
tion without continuous crack).
Even though five tests to retrieve an R&I curve seem to be little
for a correct approximation, previous experiences based on best-
fit retrieval of other experimental tests, such as reported in Borvik
et al. [6,7] and Landi et al. [8], has proved that five valid tests are
sufficient to have a good approximation of the ballistic limit vbl.
Furthermore, every nonperpendicular impact is rejected as an
invalid test. Equation (2) can be used to determine the safe ballis-
tic limit Sbl required to ensure backward compatibility with the
ISO standard for both guards, which have been tested with differ- Fig. 9 Impact test results of: (a) test 6 with sample thickness
of dpm 5 10 mm and (b) test 7 with sample thickness of
ent thicknesses dpm as in the following:
dpm 5 12 mm
Lbl10mm 57:47
cR;v10mm ¼ ¼ ¼ 1:28 (3)
Sbl10mm 45:03 these observations, two further tests (test 6 in Table 2 and test 7 in
Table 3) were carried out with initial velocities vpr,i within the
Lbl12mm 62:31 VGZ.
cR;v12mm ¼ ¼ ¼ 1:24 (4)
Sbl12mm 50:21 The test results, shown in Fig. 9, demonstrate that the projectile
penetrates the plate without passing completely to the opposite
From the given data it is possible to define a new area in the R&I side, resulting in a residual velocity vpr,r ¼ 0 m/s. However, the
diagram, called the velocity gray zone (VGZ). This Zone is guard behavior is highly uncertain and it depends on many param-
defined as the velocity interval between the safe ballistic limit Sbl eters whose slightest variation can cause different responses to the
and the lower ballistic limit Lbl for a given confidence interval. In impact.
the tests, a 90% confidence interval is used. In relation to the energy of the reduction coefficient cR,v, the
The VGZ shown in Fig. 8 is a critical zone of the velocity field term energy reduction coefficient cR,E is used analogously to gen-
where a safe impact resistance Ysf of the guard is not guaranteed erate an inference with the impact resistance Y
according to the conditions of the type-C standards. To support
L2bl10mm
cR;E10mm ¼ c2R;v10mm ¼ ¼ 1:64 (5)
S2bl10mm
L2bl12mm
cR;E12mm ¼ c2R;v12mm ¼ ¼ 1:54 (6)
S2bl12mm
In other words, the measured energy for the safe ballistic limit
Sbl is approximatively 54% lower than the energy of the lower bal-
listic limit Lbl for PC sheets with a thickness dpm ¼12 mm and
64% lower for a thickness dpm ¼ 10 mm. That means, by reducing
the kinetic energy Ekin from the lower ballistic limit Lbl through
the energy reduction coefficient cR,E you reach the safe area inside
the R&I curve, where no complete penetration occurs. This uncer-
tainty is not a fixed value but depends, for safety applications, on
material, projectile, dimension of the test pattern, and more as
already discovered by Mewes and Trapp [16].
5 Conclusion
This paper introduces a new probabilistic method for determin-
ing the impact resistance Y and the safe impact resistance Ysf for
impact tests on machinery guards in relation to the ballistic limit
vbl and the reduction coefficient cR,v. All data used to determine
the reduction coefficient cR,v for polycarbonate sheets with the
length of Lpm ¼ 300 mm, width of Wpm ¼ 300 mm, and thickness
of dpm ¼ 10 mm and dpm ¼ 12 mm are shown. The testing proce-
dure is simple and reproducible by different testing institutes if
required.
The variance of data for the impact resistance Y of safety guards
can’t be determined by using the current bisection testing proce-
dure that is based on a deterministic approach. It should be noted
that the current method has proved to be sufficient in the past by
reducing the number of injuries caused by the complete penetra-
tion of safety guards around the world. Nevertheless, by defining
the ballistic limit vbl using the R&I estimation, a possible new
standard for testing procedures is introduced. A series of impact
tests at different initial velocities vi are required to obtain a good
Fig. 8 R&I curve for a PC sheet thickness dpm 5 12 mm R&I best fit curve and a confidence interval for the ballistic limit
ASCE-ASME Journal of Risk and Uncertainty in Engineering Systems, JUNE 2022, Vol. 8 / 021001-7
Part B: Mechanical Engineering