Available online at www.sciencedirect.

com

ScienceDirect
Procedia Technology 15 (2014) 323 – 332

2nd International Conference on System-Integrated Intelligence: Challenges for Product and

Production Engineering

A strategy for logistic quality control in micro bulk production

Daniel Weimera*, Daniel Rippela, Torsten Hildebrandta, Michael Lütjena, Bernd Scholz-
Reiterb
a
BIBA - Bremer Institut für Produktion und Logistik GmbH at the University of Bremen, Hochschulring 20, 28359 Bremen, Germany
b
University of Bremen, Bibliothekstraße 1, 28359 Bremen, Germany

Abstract

A continuous trend to function integration, miniaturization and densification opens new opportunities in industry
and research. To manufacture micro products, tools, materials and technologies have to be scaled down from the
macro to the micro domain. Thereby, a downscaling of classical processes leads to unexpected process behavior, so
called size effects. Additionally, new challenges arise for in-process quality inspection based on the dimension of
the micro products which requires microscopic solutions for reliable quality control. To handle these challenges in
mass production, new strategies for the planning of logistic processes with a focus on logistic quality parameters are
necessary. This contribution introduces a closed-loop quality control strategy for bulk production in micro cold
forming. A discrete event simulation model incorporating characteristics of optical quality inspection and general
process parameters allows the quantification of the system’s performance.

© 2014
© 2014 The The Authors.
Authors. Published
Published by Elsevier
by Elsevier Ltd. Ltd. This is an open access article under the CC BY-NC-ND license (
Peer-review under responsibility of the Organizing
Peer-review under responsibility of the Organizing Committee
Committee of SysInt 2014.
of SysInt
2014.
Keywords: Micro production, quality control, metrology, uncertainty, discrete event simulation

1. Introduction

During the last decade, the demand of micro components increased strongly. Thereby, single components became
increasingly smaller while providing more functions and having more complex geometries [1]. As a consequence,

* Corresponding author. Tel.: +49-421-218-50181; fax: +49-421-218-50003.

E-mail address: wei@biba.uni-bremen.de

2212-0173 © 2014 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/3.0/).
Peer-review under responsibility of the Organizing Committee of SysInt 2014.
doi:10.1016/j.protcy.2014.09.086
32 Daniel Weimer et al. / Procedia Technology 15 (2014) 323 –
332

manufacturing of micro components became a complex task, while economic requirements increased
simultaneously. The driving force for the success of semiconductor based materials and micro-electro-mechanical
systems (MEMS) has mainly been based on advances, which allow the production of mechanics-based systems
using production systems based on electronics and semiconductors [2]. The development has further been pushed to
metals and polymers, which are not based on semi-conductor materials. To manufacture metals and polymers in the
micro domain, traditional manufacturing methods like forming have to be downscaled from the macro to the micro
domain. Based on Hansen et al. [2] components based on metals are called micro mechanical systems and the term
micro corresponds to objects smaller than 1mm in at least two geometrical dimensions [3].
One approach, to achieve high throughput rates at comparably low costs, is the application of cold forming
techniques for the micro manufacturing of metallic micro components [4]. Different cold forming processes can be
combined to achieve highly flexible manufacturing facilities with comparably low special requirements [5]. Such
production systems face the challenge of producing vast amounts of high quality components while remaining cost
efficient. Thereby, micro manufacturing is characterized by very low tolerances, the occurrence of so called size-
effects [6] and a high degree of specialized manufacturing technologies.
As a result, the adjustment of a diversity of processes in micro manufacturing poses different challenges to the
process designer. On the one hand, a careful selection and adjustment of available process technologies is required
[7]. The suitability of certain process technologies strongly depends on the manufacturing context (e.g. materials,
tools, preceding or succeeding processes). Due to the high specialization, it might even be possible that processes
have to be adapted or developed for a certain task. On the other hand, as a result of inherently low tolerances in
micro production, slight changes to a single process or material parameter can strongly impact the products quality
and therefore the overall process chain. Consequently, suitable quality inspection techniques have to be developed
and applied, in order to achieve robust and reliable manufacturing process chains. Based on quality inspection
results, process control mechanisms are applied to optimize logistic quality criteria.
This paper introduces a strategy for logistic quality control in a micro bulk manufacturing process. Different
metrologies with varying throughput and measurement uncertainty are applied in a simulation study to show their
impact on the overall production time as well as on the number of scrap parts produced.

2. Quality Inspection in Micro Manufacturing

2.1. Micro-cold Forming and Size Effects

Micro cold forming processes provide a suitable approach to manufacture high quality micro components.
Generally, such processes are characterized by high manufacturing accuracies and high throughput rates. Thereby,
the work pieces usually become hardened during the cold forming processes, which results in more robust products.
Compared to other manufacturing approaches, cold forming processes additionally lead to a reduction of waste
materials and energy consumption [8].
Although in macro manufacturing cold forming processes are well known and widely used in mass production,
they cannot be applied directly to micro manufacturing. The downscaling of those cold forming processes, and thus
of the work pieces, tools and machines, is only possible to a certain degree. Thereafter, the impact of so called size
effects impedes the further downscaling of the cold forming process [6].
Vollertsen defines size effects as “deviations from intensive or proportional extrapolated extensive values of a
process which occur when scaling the geometrical dimensions” [6]. In this context he defines intensive values as
parameters, which are not expected to change due to a change of an object’s mass (e.g. its temperature or its
density). In contrast, extensive values are expected to vary with a different mass (e.g. the object’s inertia force or its
heat content). Generally, size effects occur due to the inability to scale all relevant process parameters equally [6].
As an example, the downscaling of a metal sheet’s thickness can result in a changing density due to local defects,
although the density is considered an intensive variable. In addition to these effects, technical limitations further
facilitate the occurrence of size effects. For example, the downscaling of mechanical grippers is limited by technical
factors. For very small work pieces, the gripper’s adhesive force will eventually overcome the gravitational force at
a certain point. Consequently, the gripper will not be able to release the work piece without aid. Basically,
Vollertsen defines three distinct categories of size effects (Fig. 1) [6]:
x Density size-effects occur when the density of a material is held constant when scaling down its geometrical
dimensions. For instance, local defects become more serious with a continuing miniaturization. Thereby, the
distribution of local defects within a material can lead to more delimited sets of good and bad parts.
x Shape size-effects occur due to the increasing ratio of an object’s total surface area compared to its volume. An
example of this category is provided by the described imbalance of the adhesive force in relation the gravitational
force.
x Micro structure size-effects occur due to the fact that micro structural features (e.g. the grain size or the surface
roughness) cannot be scaled down the same way the geometrical size of an object can be.

Fig. 1. Categories of size-effects. The categories are divided into density, size and micro structure effects which lead to unexpected process
behaviour (based on [6]).

The occurrence of size effects requires a precise planning of technical parameters throughout the overall process
chain. Technical parameters between, as well as within, each manufacturing, handling and quality-inspection
process have to be regarded and adjusted to each other. Moreover, as new processes and technologies for micro
manufacturing emerge quickly, interdependencies between those parameters cannot be described precisely or are
unknown in several cases. As a result, the planning and adjustment of processes and in particular, the development
of suitable process control strategies becomes strongly dependent on highly accurate and fast in-process quality
inspection techniques. Such techniques enable the development and adaptation of suitable control strategies for
micro manufacturing for each specific manufacturing scenario.

2.2. Optical Quality Inspection and Measurement Uncertainty

An exact knowledge about geometry, forces, surface roughness and flow characteristics is required to guarantee
high quality micro products. The development of a quality control concept, integrating all of these aspects, is a
challenge and probably the reason, why quality control in the micro domain has not been completely established yet
[9]. Thereby, dimension as well as surface properties constitute major quality characteristics. The impact of
calibration uncertainties, the uncertainty of repeated measurements, uncertainty from variations of work piece
properties and the absolute value of the systematic measurement error have to be considered. In micro
manufacturing, processes are characterized by high process variability and an increased significance of measurement
uncertainty [10]. For example, the assumption that measurement devices are ten times more precise than given
tolerance intervals is hardly satisfiable in micro manufacturing. Significantly small tolerances result from absolute
small part dimensions in micro forming. In order to verify tolerances, which are necessary to ensure product
functionality, the measurement needs to be sufficiently exact. In general, measurement uncertainty remains constant,
while the tolerance conformance zone for process variations becomes smaller. The reduction of the conformance
zone in the micro dimension due to an existing and constant measurement uncertainty is shown in Fig. 2. In
addition, measured data always results of a superposition of process variation and measurement variation. Hence,
dimensional metrology is based on imprecise information, so that a probability distribution is induced.

Fig. 2. Illustration of the relationship between tolerance and measurement uncertainty (based on [11]). The measurement uncertainty is kept
constant, while the tolerance conformance zone decreases by scaling from macro to micro domain.

In general, the arbitrary shape, size and orientation of surface imperfections as well as the size of micro parts,
renders the automated surface inspection in the micro domain a challenging task. In a first step, suitable metrologies
have to be selected. Measurement techniques successfully applied for dimensional metrology in the micro domain
with respect to Hansen et al. [2] are:

x Interferometric solutions
x Microtopography measuring instruments
x Scanning electron microscopy
x Micro and Nano coordinate metrology
x Other techniques

The proposed methods are optimized to reduce measurement uncertainty and therefore to fulfill the strict
tolerance requirements for micro components. A main aspect in micro bulk production is set to in-process
capability: robustness, short measurement time and therefore high throughput rates. Tactile methods like coordinate
measuring machines offer low measurement uncertainty but are not suitable for in-process metrology and therefore
are not in focus of this contribution. A promising technique for quality inspection in the micro domain is confocal
laser microscopy (CLM). Scholz-Reiter et al. realized simultaneous 2D texture and 3D form analysis based on CLM
[12]. The CLM technique offers high resolution images and 3D data with a measurement time of 30 seconds. In a
real micro forming scenario at least hundreds of micro parts are produced each minute. Therefore, quality inspection
with a measurement time of 30 seconds requires sampling. By applying sampling, a 100% quality control of micro
components is not possible. Single defect parts cannot be detected during manufacturing. This is a crucial constraint,
especially for safety relevant parts, like micro components in airbags or control units.
There is a need for fast and accurate inline inspection metrologies in the micro range, which offer the potential to
inspect hundreds of micro parts per minute [2]. Light field cameras (LFC) show huge potential for inline quality
inspection even for dimensions smaller than 1mm. Weimer et al. introduced the LFC for 2D texture and 3D form
analysis for metallic micro components in [13]. In contrast to existing technologies like CLM, the LFC is able to
capture at least 24 frames per second. A disadvantage is the limited resolution which restricts the overall defect
detection performance. A limited resolution leads to a higher measurement uncertainty compared to more precise
technologies. Fig. 3 shows the effects of different metrologies and the specific measurement uncertainty. The green
bar represents the conformance interval based on the nominal expectation value. The uncertainty of the CLM and
LFC metrology is marked in the orange bar and reject values are marked in red. The probability distributions show
the more precise CLM metrology and the LFC with higher uncertainty (dotted distribution).

Fig. 3. Illustration of probability distributions and uncertainty observation for CLM and LFC metrologies with respect to DIN ISO 21747 [16].
The dotted distribution shows the higher uncertainty of the LFC compared to the CLM which leads to a higher rejection rate.

On the one hand a higher degree of uncertainty in the LFC case leads to a higher amount of rejections and rework.
On the other hand the LFC is able to realize a 100% inspection with a high throughput. In the following quality
control and simulation study the impact of a higher uncertainty but short measurement time is compared to a
microscopic solution.

3. Logistic Quality Control

Different quality inspection and measurement technologies are usually used as part of a larger manufacturing
system. Objective of the whole system is to produce goods of a desired quality level achieving competitive results
regarding the logistic performance of the system (such as low flow time, high due-date adherence) and ultimately
production costs. Thus, quality inspection techniques with certain accuracy characteristics and time requirements
have to be integrated in a larger quality control loop, considering quality characteristics as well as the logistic
performance of the overall system.
To evaluate the effects of different measurement technologies in the context of a micro manufacturing process,
we consider the scenario outlined in Fig. 4. We consider a single-stage micro production process followed by quality
inspection of all parts in order to ensure an overall error rate of finished goods in the low ppm (parts per million)
range. Besides these steps, we have two control functions influencing the system behavior. The first is “order
release”. For this paper we follow a constant work-in-process (WIP) strategy with a WIP level set high enough so
the bottleneck machine (which can be either production or quality inspection) never runs out of work. Furthermore,
we modeled the logistic quality control function. It receives quality measurements from the inspection step and has
to take two decisions:
First, it has to decide if a certain product should be considered as meeting the quality requirements. If not, we
have to trigger a re-production order and the part is to be considered scrap (rework is usually not an option in a
micro manufacturing process). This decision is parameterized by an accept range of the product. Using a low accept
range considers many products as being defective but those classified as good have a very high probability of
actually satisfying the quality requirements. Selecting a high value leads to a low scrap rate but bears the risk of
accepting defective goods as non-defective, so the overall quality target of a defect rate in the low ppm range is at
risk.
 The second decision of the logistic quality control function is to trigger a recalibration of the micro
manufacturing process when the process is considered out of control. This decision is based on the history of the
quality measurements and parameterized by a recalibration threshold. During recalibration the machine is not
available for processing. To minimize the time required to produce a certain amount of products, this threshold
should be as low as possible but high enough to avoid a large number of scrap parts due to the manufacturing
process being out of control.

Recalibration Request
Logistic Quality Information Flow
Control
Material Flow

Defects and Form Control Function

Deviation Fulfill
Manufacturing Quality
Order (1) Process Step
Order Targets?
Order (2)
Micro Deep Drawing [true]
.
Order Release Quality Inspection Finished Good
. Process
Order (n-1)
Order (n) [false]

Reproduction Order

Fig. 4. Outline of simulated scenario. Based on the results from the Quality Inspection step a decision based on production data whether
recalibration is needed or not is made by the Logistic Quality Control function.

4. Simulation Study

4.1. Experimental Setup

We use discrete event simulation [14] as a tool to investigate the system shown in Fig. 4 in terms of the total time
required to produce 100,000 good parts, and the total number of parts to produce to reach the target of 100,000 good
parts.
The detailed assumptions and parameter settings used in our experiments are shown in Table 1. The process
settings resemble the manufacturing process of a micro cup with a desired diameter of 500μm. This diameter can
deviate at most by 10μm for the micro cup still to be considered good. We model the manufacturing process to
create parts with a diameter following a normal distribution with a mean of 500μm and mean standard deviation of
6.67μm. These figures are valid for a machine just calibrated. To model tool wear and other sources leading the
process to get out of control we slightly increase the mean of this distribution each time a new part was produced.
This increase leads to about 50% of all parts produced to be scrap after 43,200 parts being produced (i.e., after three
hours of full-speed production the mean is increased by 10μm).
 Table 1: Parameter setup for simulation study.
Process Parameters Value/Range
Width N(μ=500μm, std=6.67μm)
Mean shift 10μm after 3h full-speed production
Time / part 0.25s
Recalibration time 1h
Metrology I (Lightfield Camera)
Uncertainty ±3μm (±3std)
Measurement time 0.1s
Metrology II (Confocal Laser Microscope)
Uncertainty ±0.8μm (±3std)
Measurement time 30s
Experimental Parameters
Parts to make
100,000 pcs
Quality Target
500μm ± 10μm
Accepted range ± {4, 5, 6, 7, 8, 9, 10}μm
Recalibration Threshold {3, 4, 5, 10, 15, 20, 25}pcs

We consider two different measurement technologies characterized by a certain measurement uncertainty

(characterized by a normally distributed measurement noise) and their time requirements. The first technology
(resembling characteristics of a light field camera) has a rather large measurement uncertainty of ±3μm but can
make up to 10 measurements per second. Using this measurement technology, the manufacturing process can
operate at full speed and constitutes the bottleneck resource of the system. The second technology (resembling a
CLM) is considerably more accurate (±0.8μm), but requires a measurement time of 30s. Therefore using this
technology, the quality inspection step is the bottleneck resource of the system. Besides these two technologies we
also simulate the effects of a hypothetical, perfect measurement technology with a measurement uncertainty of 0 in
order to quantify the loss in terms of production time and produced quantity due to a realistic measurement
uncertainty.
In order to determine suitable configurations for the logistic quality control, we consider seven different values
for the accept range as well as seven different values for the recalibration threshold for these four technology
variants (measurement technology 1, measurement technology 2, hypothetical perfect technology - fast, hypothetical
perfect technology - slow). We use a full factorial experimental design to investigate all possible parameter
combinations leading to 7×7×4=196 simulation runs. Each run consists of 30 independent replications, each
simulating the system until 100,000 good parts were completed. The simulation was implemented using the discrete
event simulation library jasima ([15], Java Simulator for Manufacturing and Logistics,
http://jasima.googlecode.com).

4.2. Experimental Results and Discussion

Analyzing the results in terms of the required time and the total number of parts produced we can identify a small
set of non-dominated parameter settings. This means, all of these settings offer different possible trade-offs between
time and number of produced goods. All other parameter combinations are dominated by these settings, i.e. they
lead to larger processing times without requiring fewer parts to be produced than all of the non-dominated settings.
Detailed results from all simulation runs are shown in Table 2 and graphically for the fast measurement
technologies in Fig. 5. In this case the manufacturing machine is the bottleneck machine and therefore there are
different trade-offs between the number of parts to be produced and times without production due to recalibration.
Smaller values for the recalibration threshold lead to a higher number of recalibrations and an increase in the total
production time, but on the other hand lead to fewer scrap parts caused by an un-calibrated machine. Which
combination of parameter settings will finally be implemented depends on the preferences of a user and probably on
the costs associated with production time and the variable costs associated with each part. Without knowing such
preferences (6,10) (Accepted Range, Recalib. Thresh.) for measurement technology 1 and (10,5) for the
hypothetical, perfect technology seem to be good compromises. Moving in either direction form one of these points
requires a proportionally large increase in one objective leading to only a slight improvement in the other.
Comparing these settings for the hypothetical perfect and real measurement technology we can quantify the loss due
to a rather imprecise measurement technology to an increase of about 32% in production time and an even larger
increase of about 41% in the number parts to produce. The figures for the hypothetical technology also offer a lower
bound for all measurement technologies that can cope with the maximum speed of the manufacturing machine. No
matter how accurate they are, we can’t get better results, but only hope to get to its results as close as possible (of
course only for the process characteristics as given in Table 1).

Table 2: Detailed experimental results of all non-dominated parameter combinations. Values in brackets show twice the standard error over
the 30 independent replications performed.

Technology Accept Range Recalib.Thresh. Time [h] Produced [pcs] Recalib.Count

hypoth.-fast 10 3 244.4 (±5.7) 115,411.2 (±46.4) 236.4 (±5.7)
hypoth.-fast 10 4 42.9 (±2.1) 115,841.6 (±47.7) 34.9 (±2.1)
hypoth.-fast 10 5 18.2 (±0.7) 117,684.5 (±190.2) 10.0 (±0.8)
hypoth.-fast 10 10 12.1 (±0.0) 131,355.0 (±644.6) 3.0 (±0.0)
hypoth.-fast 10 15 12.0 (±0.1) 141,389.2 (±583.5) 2.1 (±0.1)

MeasTech1 6 5 734.7 (±12.3) 159,464.1 (±137.8) 723.7 (±12.3)

MeasTech1 6 10 24.0 (±0.6) 166,221.4 (±566.8) 12.4 (±0.6)
MeasTech1 6 15 18.2 (±0.2) 183,039.1 (±714.1) 5.5 (±0.2)

hypoth.-slow 10 3 962.7 (±0.4) 115,410.6 (±45.9) 191.7 (±3.8)

MeasTech2 9 3 1014.1 (±0.6) 121,510.8 (±62.7) 363.1 (±5.7)

For the two slow technologies (measurement technology 2 and hypothetical-slow) there is no such trade-off
between time and the number of parts to produce (see Table 2). Using these settings, the measurement technology is
the bottleneck resource and the manufacturing machine has enough free capacity to frequently recalibrate. Together
with the fact that measurement technology 2 is pretty accurate, and we can thus use a large acceptance range, which
leads to a slight increase of about 5.2% in the number of parts to be produced and an increase in production time
also of about 5.2%.
 190000 (6,15)

Produced [pcs]
170000 (6,10)
(6,

150000
(10,15)
(10,10)

130000
(10,5) (10,4) (10,3)

110000
10 100
Time [h] (log-scale)

Fig. 5. Possible trade-offs between total number of parts produced to achieve 100,000 good parts and the total time required to do so. Circles
show the results of a hypothetical, fast, and perfect measurement system, triangles show the results obtained with “Measurement Technology
1”.

5. Conclusion and Future Work

This article investigated the influence of different in-process quality inspection techniques by means of a material
flow simulation. Thereby, the effects of a slow but highly accurate CLM are compared to a less accurate but faster
LFC, each with respect to the overall production time as well as to the amount of scrap parts produced. Comparing
these technologies leads to different trade-offs in time vs. parts produced. The fast but imprecise technology requires
an additional overhead of about 37% parts to produce; however requiring only 1/42 of the time. Given usually low
variable costs of micro-manufactured parts, this seems to be the better choice, but a more detailed analysis would
require a cost model evaluating the overall production costs over a long time. This result however is also interesting
as it stresses the usefulness to simulate the complete system and not just look at the characteristics of the
measurement technology. An increase in in production time by a factor of 42 is clearly worse, but looking at the
measurement times alone we would expect much larger values (of about 120 = 0.25/30). Tuning the quality control
strategy appropriately to the characteristics of the manufacturing process and the characteristics of the quality
inspection using discrete event simulation allows more accurate results than just looking at them in isolation. In
future work the proposed logistic quality control strategy will be adapted to a demonstration platform which is
described in [13]. This platform combines material transport, production, abrasive tool wear, sorting, bulk handling
and quality inspection processes. To overcome the limits of a single measurement system a combination of different
metrologies will further improve the overall quality inspection reliability and will be considered in future research.

Acknowledgements

The authors gratefully acknowledge the financial support by Deutsche Forschungsgemeinschaft (DFG, German
Research Foundation) for Subproject B5 “Reliable Processes” and Subproject C4 “Simultaneous Engineering”
within the CRC 747 (Collaborative Research Center).

References

[1] Wulfsberg JP, Redlich T, Kohrs P. Square Foot Manufacturing: a new production concept for micro manufacturing. Production Engineering
- Research and Development. 2010;4(1):75 - 83.
[2] Hansen HN, Carneiro K, Haitjema H, De Chiffre L. Dimensional Micro and Nano Technology. Annals of the CIRP. 2006;55[2]:721-734.
[3] Geiger M, Kleine M, Eckstein R, Tieslerl N, Engel U. Microforming. CIRP Annals - Manufacturing Technology. 2001;50[2]:445-462.
[4] Qin Y. Micro-forming and miniature manufacturing systems - development needs and perspectives. Journal of Materials Processing
Technology 2006;177(1-3):8 - 18.
[5] Klemd O. Desktop Factory – New approaches for lean micro assembly. Proceedings of the IEEE International Symposium on Assembly
and Manufacturing 2007 (ISAM '07); Ann Arbor, MI2007. p. 161 – 165.
[6] Vollertsen F. Categories of size effects. Production Engineering. 2008;2[4]:377-383.
[7] Scholz-Reiter B, Lütjen M, Brenner N. Technologieinduzierte Wirkungszusammenhänge in der Mikroproduktion - Entwicklung eines
Modellierungskonzepts. In: Schenk M, editor. 22 HAB-Forschungsseminar „Digital Engineering - Herausforderung für die Arbeits- und
Betriebsorganisation“. Magdeburg: GITO Verlag; 2009. p. 81-102.
[8] DeGarmo EP, Black JT, Kohser RA. Materials and Processes in Manufacturing. 9th ed: Wiley; 2003.
[9] Pfeifer T, Driessen S, Dussler G. Process observation for the assembly of hybrid micro systems. Microsystems Technologies 2004;10:211-
218.
[10] Fleischer J, Lanza G, Schlipf M. Statistical quality control in micro-manufacturing through multivariate ȝ-EWMA chart. CIRP Annals -
Manufacturing Technology. 2008;57:521-524.
[11] Tosello G, Gasparin S. Applications of dimensional micro metrology to the product and process quality control in manufacturing of
precision polymer micro components. CIRP Annals - Manufacturing Technology. 2009;58:467-472.
[12] Scholz-Reiter B, Weimer D, Thamer H. Automated Surface Inspection of Cold Formed Micro Parts. CIRP Annals 2012 - Manufacturing
Technology, Annals of the International Academy for Production Engineering. 20012;61[1]:531 - 534.
[13] Weimer D, Thamer H, Fellmann C, Lütjen M, Thoben K-D, Scholz-Reiter B. Towards 100% in situ 2D/3D quality inspection of metallic
micro components using plenoptic cameras. Procedia CIRP (in print).
[14] Law AM. Simulation Modeling and Analysis. 4th Edition ed. New York, USA: McGraw-Hill; 2007.
[15] Hildebrandt T, Scholz-Reiter B. Optimierte Steuerung von Logistikprozessen - Simulationsbasierte Optimierung auf Basis leistungsfähiger
Open Source Simulation. Productivity Management. 2013;18(5):53 - 56.
[16] Norm ISO 21747: Statistical Methods – Process Performance and Capability Statistics for Measured Quality Characteristics. 2006.