Page 310 -

292 11 Analyzing “Lasagna Processes”

or number of production cycles). Moreover, there may be Service Level Agree-
ments (SLAs) specifying a fine if the machine is down for an extended period.
Event data can be used as a basis for billing and checking SLAs. Moreover, the
manufacturer gets insights into the way that machines are used, when they mal-
function, and when they require maintenance.

These examples show that there are opportunities for process mining in all three
economic sectors.

11.4.3 Two Lasagna Processes

To conclude this chapter, we briefly discuss two case studies analyzing Lasagna
processes.

11.4.3.1 RWS Process

The Dutch national public works department, called “Rijkswaterstaat” (RWS), has
twelve provincial offices. We analyzed the handling of invoices in one of these of-
fices [106]. The office employs about 1000 civil servants and is primarily respon-
sible for the construction and maintenance of the road and water infrastructure in
its province. To perform its functions, the RWS office subcontracts various par-
ties such as road construction companies, cleaning companies, and environmental
bureaus. Also, it purchases services and products to support its construction, main-
tenance, and administrative activities. The reason to employ process mining within
RWS was twofold. First of all, RWS was involved in our longitudinal study into
the effectiveness of WFM systems [76]. In the context of this study, RWS was
interested to see the effects of WFM technology on flow times, response times,
service levels, utilization, etc. Second, RWS was interested in better meeting dead-
lines with respect to the payment of invoices. Payment should take place within 31
days from the moment the invoice is received. After this period, the creditor is en-
titled (according to Dutch law) to receive interest over the outstanding sum. RWS
would like to pay at least 90% of its invoices within 31 days. However, analysis of
the event logs of RWS showed that initially only 70% of payments were paid in
time.
Starting point for the analysis described in [106] was an event log containing in-
formation about 14,279 cases (i.e., invoices) generating 147,579 events. Figure 11.8
shows a C-net generated by the heuristic miner. This model shows that the RWS
process is fairly structured, but not as structured as the WMO process depicted in
Fig. 11.2(a). After some efforts (filtering the log and tuning the parameters of the
mining algorithm), it is possible to create a model with a fitness of more than 0.9.
The log can be replayed on this model to highlight bottlenecks. Such analysis shows
that several activities had to be redone (as can be seen by the loops of length one or