APPLICATION OF OPERATIONS SCHEDULING ALGORITHMS FOR THE ALLOCATION OF FORENSIC SCIENCE SERVICES: A CASE STUDY

: The activities of preparing and issuing Forensic Science Services reports are a public service of great importance for society and for the executive and judicial branches. However, notably, it has difficulties in meeting due dates, causing damage to all involved and generating additional costs for governments. In practice, it is observed that knowledge in the area of operations management is not used to determine the allocations of Forensic Science services at Criminalistics Technical Sections. Therefore, the objective of this research is to develop and apply knowledge related to the operations scheduling of Forensic Science reports. To fulfill the objective, the case study research method was applied in the Regional Technical Section of Criminalistics (STRC) of Três Corações - Minas Gerais in Brazil. Real data from that Forensic Unity carried out in the past were implemented in an APS (Advanced Planning and Scheduling) software and new allocations, based on the software's internal algorithms, were carried out. The results of allocations from the software have been compared with the previous results. The comparisons show that scheduling from the software's algorithms can bring better results.


INTRODUCTION
Among the public services that most influence people's daily lives are those of public safety and criminal justice.The lack of this service or its poor provision can bring a lot of damage to society.For example, Freedy et al (1994) report that crime victims involved in the criminal justice system are at risk for developing posttraumatic stress disorder.One of the agents of this public service chain is the Forensic Science Service (FSS).The FSS is the agency in charge of producing scientific material evidence, which transformation resources are scientific knowledge and the technologies applied to process the traces found at crime scenes.The FSS is essential in reducing crime because it increases the chances of a successful investigation into crimes and a fair trial.Forensic evidence has scientific validity and, at the same time, helps in the promotion of human rights, preventing suspects from being subjected to physical and psychological constraints.
An event that generates the need for a criminal investigation represents the demand for this public service and the FSS report with the forensic scientists' conclusions represents the delivery, that is, the fulfillment of the demand.One of the first decisions to be made in this process is the allocation of the task of preparing the FSS report to one of the forensic scientists on duty at Criminalistics Technical Section.Notably, the allocation of resources to perform tasks is the first decision in operations scheduling problems (Baker, 1974).At FSS unities, such allocation is made based on the experience of the head of the Criminalistics Technical Section.
Simplifications like this in decision making in operations scheduling are common in practice, mainly due to the complexity inherent to these problems, however the results obtained in such situations are not necessarily satisfactory (Carvalho et al., 2014).
The FSS has complex management (Koppl, 2005, P. 256;Brasil, 2006) and several opportunities for improvement.One of the most urgent opportunities for improvement in the provision of the FSS is to meet the due dates (Rodrigues;Toledo, 2015a;Rodrigues;Toledo, 2015b;Belluco;Pimenta, 2013;Gonçalves, 2013;Lima;Goldszmidt, 2013;Houck et al. 2012;Brasil, 2012, p. 101-104).An example is São Paulo, whose fact was reported in the national press (Soares, 2015;Zúnica, 2014), but it can be extended to other units of the federation, as shown by the document of the Ministry of Justice in Brazil (Brazil, 2012).In part, such delays can be attributed to the lack of methods for allocating the tasks.
Although there is a need to apply operations scheduling knowledge in such services, recent literature shows that there are still gaps to be filled.Research on scheduling has been carried out based on various topics such as I4.0 (Ghaleb et al., 2020;Parente et al., 2020;Hu et al., 2020) and supply chain management (Jamrus et al., 2020;Solina and Mirabelli, 2021;Chavez et al., 2020), but little research deals with scheduling in services.In a survey considering case studies in operations scheduling, Fuchigami and Rangel (2018) found that there are no articles in the literature that address service companies.Based on a literature review, Carvalho et al. (2014) argue that there is a gap in research related to operations scheduling when it comes to applications in real contexts.More recently, Parente et al. (2020) also conclude that more research is needed on operations scheduling problems.
The present work aims to fill some of the identified gaps.First, the research detailed in this paper presents a case study carried out in a service organization.
More than that, the case study was carried out in a public agency, in a FSS, which, according to the authors' knowledge, is unprecedented in the literature related to operations scheduling.Thirdly, the results obtained in this research offer opportunities to improve the provision of this service to clients (judges, prosecutors, defense attorneys, police investigators, etc.) and other stakeholders, such as victims, suspects, media and political agents.
The objective of this research is, using real data from a Forensic Science Unity, to apply algorithms of an Advanced Planning and Scheduling software and to compare the results obtained by the software with the results and scheduling obtained previously (without the use of the software).When defining this objective, the researchers supported themselves in the hypothesis that APS systems greatly assist the sequencing and dispatch of orders offering good results, as shown in Ivert (2012).
The structure of this paper is as follows: in sections 2 and 3 the theoretical references of the study are presented; in section 4 the research method are explained; in the section 5 the case study and results are described in detail; and, finally, in section 6 the final considerations are made.

PUBLIC SAFETY AND CRIMINAL JUSTICE SERVICES CHAIN AND FORENSIC SCIENCE OPERATIONS
The FSS is part of a supply chain of public safety and criminal justice services.
This chain must deliver a justice value to the society in general, and the final product is the condemnatory or absolute judicial sentence.
According to the law, the ostensive or patrol police (Military or Highway Police), upon learning of a fact, must go to the scene, isolate it and call the Investigative Police (Civilian Police), who will take over the case, while the ostensive Police return to patrolling.If there are traces any at the crime scene, the police investigator will request the Forensic Science services and preserve the crime scene until the forensic scientist (in this case, a Crime Scene Investigator -CSI) arrives, immediately starting the investigation (Brasil, 1941, Art. 6).
The legislation determines the obligation to carry out the forensic science examination in every criminal infraction that leaves traces, under penalty of nullity of the criminal process.These examinations must be carried out by a CSI (BRASIL, 1941, arts. 158-159).The forensic scientists enjoy technical, scientific and functional autonomy (Brazil, 2009) and are considered auxiliary to the Justice System.
The FSS is complex, as it has interfaces with the technical-scientific, police and legal spheres (Misse et al., 2006).It is a professional service (Mintzberg, 2009, P. 212;Silvestro, 1999), where, in many cases, the forensic scientist must apply to the specific case the knowledge learned during the university course -engineering, medicine, dentistry, pharmacy, physics, chemistry, biology, geology, computing, accounting sciences etc. -and in the professional training course (Rodrigues, 2010, p. 124).
The service process begins with triggering, as a rule, by the Civilian Police.
When the crime scene is external (homicides, work accidents, robberies, kidnappings etc.), the CSI moves and transports all the necessary material to the crime scene.On site, the CSI draws sketches, takes photographs, takes measurements, documents and collects materials for complementary examinations (Rodrigues, 2010;Rodrigues et al., 2010).
After the crime scene exams, the CSI returns to the FSS Office and, if necessary, requests laboratory exams and / or other specialized exams from the back office of the service, such as DNA, ballistic, toxicological, chromatography, among others.At the back office, the service is organized functionally into specialized units such as forensic biology, legal physics, legal chemistry, forensic ballistics, for example.Subsequently, the forensic scientist prepares the report, which contains the examinations and conclusions, and send it to the client who requested it (Rodrigues, 2010).
In the judicial phase, the forensic scientists may be subpoenaed by the judge to answer questions (questions about points of the report) in writing or personally attend the hearing at the criminal trial to provide clarification on the FSS report in a specific case.The frontline operations operate 24 hours a day, seven days a week (Rodrigues;Rachid, 2006;Rodrigues, 2010;Rodrigues et al., 2010).
Notably, the FSS presents several opportunities for improvement.One of the most urgent is the dependability, as evidenced by Soares (2015) and Zúnica (2014) and as shown by the document of the Ministry of Justice for the diagnosis in Brazil (Brazil, 2012).The delay in making available FSS reports harms the police investigatons, public prosecutors in the criminal courts suits, defense attorneys in defining the best defense strategy and judges in their sentences, thus making value delivery more difficult.
The main problems related to the delivery of reports are the scheduling of services (examinations and reports) and the finite capacity on personnel, buth forensic scientists and clerical workers.

THE CONTRIBUTION OF THE OPERATIONS SCHEDULING TO THE FSS
When orders arrive in a production or service system, a decision must be made on the order in which the tasks will be performed.This priority is often established by simple rules, for example, First-Come, First-Served (FFSS) and Earliest Due Date (EDD), both widely used in practice.However, for most situations, these simple rules do not provide good results (Slack et al., 2009;Chase et al., 2006).
Production scheduling occurs at four levels in the planning process; aggregate scheduling, and product completions (MPS), and the sequencing of jobs on machines (Buxey, 1989).According to Morton and Pentico (1993), operations scheduling is the process of organizing, choosing and timing the use of resources to carry out the activities necessary to produce the desired outputs at the desired times, while satisfying a series of time restrictions and relationships between activities and resources.This definition calls attention to three important points: resources are used, the outputs are related to objectives and there are restrictions.Notably, for any system, resources are always limited.Still, the system needs to achieve its objectives, for example, serving customers on time.And any and all tasks must be performed according to their productive route, that is, there are stages to be fulfilled and that are interdependent.Given these characteristics, operations scheduling is an complex activity, because if only the number of different sequences to perform n tasks is considered, the difficulty is already great, since they are n! possibilities to be analyzed for decision making.
Another difficulty in operations scheduling is keeping the schedule up to date.This is because unforeseen events occur in the production systems: machines break down, workers are missing, suppliers delay deliveries, etc.Thus, the recently planned programming, even with very well elaborated methods, loses its effectiveness in a short time.In other words, in unstable environments, in order for the objectives to be achieved, the production schedule must be redone with a very high frequency.
Due to these difficulties in solving operations scheduling problems, on the one hand, many researches are focused on the development of heuristic algorithms for their realization.On the other hand, many of these heuristic algorithms are included and developed to work in software that allows to do the scheduling and update it in a fast and friendly way.Such software is known as Advanced Planning and Scheduling Systems (APS).As highlighted by Ivert (2012), APS systems make the scheduling algorithms practical and more attractive.
The carrying out of criminal investigations and the preparation of forensic reports is precisely a situation in which the use of an APS software can offer a contribution.You never know when a crime will happen.The deadlines for delivering the reports vary from case to case.The length of time a forensic scientist can perform can vary widely.Forensic personnel work on call.The number and variety of different reports to be allocated to each forensic scientist is very large.And lives are involved in this whole process (victims, accused, innocent, etc.).Although there is all this complexity, in practice the operations scheduling in FSS is done based only on priorities defined by the heads of the sections, manually or using electronic spreadsheets, without knowing if the scheduling will really bring better results in terms of dependability and better capacity utilization.That is, although there are improved management tools, in practice it is common to use methods that are not aligned with the evolution of knowledge (Bernardo et al., 2022).With this, what happens are delays in the delivery of the reports and, consequently, a negative effect in the entire criminal investigation and criminal justice system, waste of public resources, in addition to the risk of violations of the human rights of those involved.
As Naves (2016) argues, the history of the Brazilian criminal area is marked by the distance between the academy and the knowledge construction process, which, in part, makes it difficult to validate scientific techniques, procedures and conclusions.Even though there is a noticeable gap, authors like Sala (2018) are emphatic when reporting that "at the institutional level, the most effective infrastructures must be developed in all areas of research, practice and politics".This is also corroborated by Santos et al. (2016) who state that "in the provision of public services, the growing demands of citizens to meet their needs and the need to rationalize public resources have forced governments through their public administration bodies to act more and more efficiently".Still, according to the same authors, FSS is one of the sectors with little attention paid by researchers in the area of operations management (Santos et al., 2016), although it is a tool in the service of justice that uses science in favor of clarifying facts (Silva, 2019).
One of the improvement opportunities identified by the authors of the present work is to employ sequencing rules or operations scheduling algorithms for the allocation and realization of Forensic Science activities.

RESEARCH METHOD
The objective of the research is to apply algorithms of an APS software and to compare the results obtained by the software with the results and scheduling obtained previously.
To achieve the proposed objective, the case study method was applied.The case study is appropriate for the present study since it is a way of doing empirical social research, with the objective of investigating the phenomenon within its context.
For Yin (1990), case studies are suitable for opportunities such as those in the present study.For this research, we chose to use the case study as a means of exploring the results that would be obtained in a practical situation in which the peculiarities are complex and, in addition, there was no clarity about the results that would be achieved.Among the disadvantages of the case study are the difficulty of generalization and the subjective nature of the measurement process.In the present study, however, it is not intended to generalize the results; only, from the primary sources, understand the phenomenon through the perspective of the object of study.
In order to circumvent the subjective nature of the measurement of results, only analyzes of the reports obtained directly from the software were performed.
The composition of the case study was based on the propositions of Yin (1990), Meredith (1998) and Miguel (2007).First, a theoretical framework was built based on a literature review.Then the research question was elaborated.With regard to the guiding propositions, bearing in mind that these should help to focus attention on essential data (Yin, 1990), it was determined that the historical data of entry, allocation and deliveries of the FSS reports of a criminalistics technical section.
With the previous proposition, the definition of the unit of analysis is clearly determined.Finally, the researchers decided not to elaborate a data collection protocol, as the necessary data would be collected directly from the information system of the Forensic Science unity.
All information related to the forensic science activities was obtained directly from the Regional Technical Section of Criminalistics, located at Avenida Deputado Renato Azeredo, n. 1342, in the city of Três Corações, Minas Gerais state, Brazil.
The sources of information were basically spreadsheets stored on the unit's computers and also in written records in books.The information is: It is important to highlight that the 2015 information was selected because it is the most recent available in spreadsheet format.Information from later years (from 2016 on) was recorded only in the official program of the Civilian Police of Minas Gerais, called PCNET.This system is a typical transactional system for recording requests for reports, codes, nature of exams, requesting units, dates (requisition, acceptance, forwarding, dispatch, etc.), assigned forensic scientist etc.
The research was carried out in three steps.The first step consisted of applying the sequencing rules available in the Preactor software using the reference times established by the Superintendence of Technical-Scientific Police of the Civilian Police from Minas Gerais state.The standard times for FSS reports are in the Annex A and details of the rules are described in the case study section.Sequencing results and resulting times were obtained from the software.With this step it was possible to verify the best rules for the problem studied.The second step was to apply only the rules with the best results from the previous step, now using the real times of the reports performed.The real times were obtained from the information system of the criminalistic technical section.In the third step, both real times and forensic scientist allocations carried out during the analyzed period were modeled in the software.From this modeling, the same previous sequencing rules (steps 2 and 3) were applied.In each of the steps, the average completion times of the activities, the forensic scientist allocations carried out (similarities and differences between rules and between the rules and real allocations) and the total number of reports carried out were analyzed.Step 1 application of sequencing rules using reference times Step 2 application of the best performing rules using real times Step 3 application of the best performing rules using real times e real allocations identification of the best performing rules comparison and analysis of results

comparison and analysis of results
Source: Authors.
The replication of this research requires that the same software and version be used so that the results are compatible.

Encoding of information in the software
The information obtained was encoded in the Preactor® software, as follows: -the forensic scientists were registered as primary resources with the following properties: "Finite Mode Behavior" Infinite with Shift, "Behavior in Infinite Mode" Infinite and "Efficiency" 100%.The forensic scientists were registered as infinite resources because the number of reports attributed to each of them cannot be limited, which, according to the data collected, is confirmed by the large number of reports attributed to the same forensic scientist over time; -the FSS reports were registered as products without operating times.It was decided not to allocate operating times in the register so that when the reports was included as demands (orders in the software) it was possible to input the time according to the purpose of the analysis.To analyze the real results, the real times were included and to analyze the theoretical results, the standard times were include.
Table 1 below summarizes the information from the registered FSS reports; -the reference times (standards) for carrying out FSS reports were used to include the order times when analyzing the theoretical results in "Operation Times ...", "Time per item"; -shifts were registered as a "Primary Calendar Standard" and assigned to each forensic scientist; -the forensic reports carried out were registered as orders with the following properties: In "Order No." the number of the report was informed, in "Code" and "Product" the respective FSS report and in "Earlier Start Date" the date of entry of the service demand.For forensic examinations with an entry date prior to the month of March, the first day of March of two thousand and fifteen was assigned as the earliest start date.In "Resources ..." and "Times of Operation ..." the data were inserted according to the analysis that was intended to be carried out.The remaining properties were maintained according to the software's default as they did not interfere with the analyzes to be performed.
The researchers chose the month of March 2015 to carry out the study because it is the month with the highest total number of recorded services.

Application of software and verification of solutions
The researchers used the software to assess how forensic scientists reports would be scheduled using the algorithms present in the software.Two hundred and nine FSS reports were registered (Appendix 1).

FSS reports scheduling using reference times
First, it was verified how the forensic scientists' schedule would made in the month of March 2015 through the use of the software using the reference times And the APS Preferred Sequence rule sequences operations based on attributes (color, flavor, package size, product family) of the products or the operations themselves.For each of the rules used, the results were obtained through the statistics of the schedules and reports of the schedules.

FSS reports scheduling using real times
After testing the different rules, it was concluded that the rules with the best overall results were the APS Forwards and APS Backwards.So, the next step was to see how the results would look if the same rules were applied and the times were the same as the real times (and no longer the reference times).For this, the forensic reports were registered as orders with the same start dates and the processing times equal to the real times.
In a first attempt to enter data, carried out only by changing the times in the schedule itself, that is, with the schedule already defined by Preactor®, the times of each order were changed one by one.As a result, many forensic reports was no longer allocated due to the lack of resources (insufficient capacity).For each order that was no longer allocated, the software presented the following warning: "Alert: Record "X" does not adjust to the available time of Resource "Y"".For this reason, after the end of the modification of the times (exchange of reference times for real times), all the forensic reports were deprogrammed and then reprogrammed with the same previous rule (in this first case, the APS Forwards rule).Statistics and result reports were collected and it was found that there were changes in relation to the initial schedule.
As it was found that the insertion of real times in the previous schedule (programming with reference times) causes the deprogramming of the forensic reports due to lack of capacity (which was verified in the case of the APS Forwards rule), for the APS Backwards rule the insertion of real times was performed prior to scheduling.Statistics and result reports were collected and it was found that there were changes in relation to the initial schedule.

Real forensic reports scheduling
The third application of the software and verification of the solutions was carried out as follows: in Preactor®, the real allocation for the month of March 2015 was implemented.The FSS reports were registered as orders with the same start dates (in the software defined by the earlier start dates) and the same allocation of forensic scientists (in the software defined by the required resource).In addition, the processing times were recorded according to the real times (in the software defined by the time of operation, time per item, in the order window).To generate the schedule, the APS Forwards and APS Backwards rules were used.Statistics and reports were collected.

Comparison of rules using reference times
According to the statistics collected, the APS Forwards, APS Parallel Loading and APS Preferred Sequence rules generated very similar results.The APS Backwards rule generated slightly different results.In terms of the quality of allocations, there was a technical tie between the APS Forwards, APS Parallel Loading and APS Preferred Sequence rules and a small advantage of the APS Backwards rule over the others because the number of incomplete activities was less.
In an electronic spreadsheet, the differences between the end dates and the start dates of each FSS report were calculated.Then, the average values were calculated for each rule.The APS Forwards, APS Backwards, APS Parallel Loading and APS Preferred Sequence rules resulted in average times equal to 9:54:26, 10:46:30, 14:31:11 and 13:21:40 (hour:minute:second), respectively.
Each rule was compared in terms of forensic scientist allocation for each report.Table 2 summarizes the similarities between forensic scientist allocations in relation to each pair of rules.

Comparison of rules using real time
According to the statistics collected, the APS Forwards rule with real times resulted in 20 reports unallocated while the APS Backwards rule resulted in only one.
The other statistical results are similar or with differences that do not represent superiority or inferiority of one in relation to the other.
In an electronic spreadsheet, the differences between the end dates and the start dates of each report were calculated.Then, the average values were calculated for each rule.The APS Forwards and APS Backwards rules resulted in average times equal to 93:46:11 and 105:07:36, respectively.

Comparison of rules using real times and real allocation
As per the statistics collected, the APS Forwards rule with real allocation resulted in 25 reports not being allocated while the APS Backwards rule with real allocation resulted in 3 unallocated.The other statistical results are similar or with differences that do not represent superiority or inferiority of one in relation to the other.
In an electronic spreadsheet, the differences between the end dates and the start dates of each report were calculated.Then, the average values were calculated for each rule.The APS Forwards and APS Backwards rules with real times resulted in average times equal to 93:46:11 and 105:07:36, respectively.The APS Forwards and APS Backwards rules with real allocation resulted in average times equal to 101:50:03 and 124:42:05, respectively.
In a spreadsheet, the differences between the end dates and the earlier start dates of each report were calculated, that is, the flow times.Then, the mean values for each rule were calculated, that is, the mean flow times.For this comparison, only the reports that were allocated by all rules was considered.Under these conditions, the APS Forwards rule with real times, APS Forwards with real allocations, APS Backwards with real times, APS Backwards with real allocations resulted in average flow times equal to 91:27:42, 114:55:03, 457:56:04 and 419:28:31.The real allocation, considering the same reports, has an average flow time of 101:55:54.
Table 3 summarizes the similarities between FSS reports allocations to forensic scientists in relation to each pair of rules.

Summary of comparisons between rules
Using the reference times, the APS Forwards and APS Backwards rules presented the best results in terms of scheduling statistics and scheduling reports generated by the software itself and also in relation to the average times of the differences between the end dates and the start dates of each report.As a result, these two rules were used to make comparisons between the results that would be obtained with the real times.
Using real time, the APS Forwards rule resulted in a greater number of unallocated reports due to lack of capacity (9.57% non-allocation) compared to the APS Backwards rule (0.48% non-allocation).Regarding the average times of the differences between the end dates and the start dates of each report, the APS Forwards rule resulted in a lower value, equivalent to about 90% of the average time of the APS Backwards rule.
The reports were also inserted in the software with the real allocation, that is, in addition to the real times, the reports were allocated to the forensic personnel according to the historical record.For these allocations, the APS Forwards rule resulted in a greater number of unallocated reports due to lack of capacity (11.96% non-allocation) compared to the APS Backwards rule (1.44% non-allocation).
Regarding the average times of the differences between the end dates and the start dates of each report, the APS Forwards rule resulted in a lower value, equivalent to about 82% of the average time of the APS Backwards rule.Comparing the average flow times, the only rule with a lower result in relation to the real allocation was the APS Forwards rule, with about 90% of the average flow time of the latter.The graph in figure 2 summarizes the average flow times of the rules.Regarding the comparison between the rules using reference times with respect to the allocation of reports to forensic personnel, the results showed a similarity of 5.7% between the rules APS Forwards and APS Backwards, 78.9% between the rules APS Forwards and APS Parallel Loading, 93.8% between APS Forwards and APS Preferred Sequence rules, 7.7% between APS Backwards and APS Parallel Loading rules, 5.3% between APS Backwards and APS Preferred Sequence rules and 84.7% between rules APS Parallel Loading and APS Preferred Sequence.
Regarding the comparison between the rules using real times with respect to the allocation of reports to forensic scientists, the results showed a similarity of 14.83% between the rules APS Forwards and APS Backwards.
Regarding the comparison between the rules using real times and the real allocation with respect to the allocation of reports to forensic scientists, the results showed a similarity of 26.3% between the APS Forwards rule with real times and APS Forwards with real allocation, 22% between APS Backwards rule with real time and APS Backwards with real allocation, 27.3% between APS Forwards rule with real time and APS Backwards with real allocation, 15.3% between APS Backwards rule with real time and APS Forwards with real allocation.

Conclusions on comparisons between rules and discussion
From the results obtained, it can be concluded that the scheduling performed from the software's internal algorithms are, at the same time, different and better than the real scheduling performed manually by the head of the section.The similarities obtained from the algorithms in relation to the real allocation ranged between 15.3% and 27.3%.The best result obtained by the algorithm in terms of average flow time showed a reduction of about 10% in relation to the real scheduling.
Regarding the scheduling carried out by the software, it is concluded that the APS Forwards and APS Backwards rules were the ones that presented the best results (in relation to the results given by the APS Parallel Loading and APS Preferred Sequence rules).
In the day-to-day practice of the sector, scheduling is performed manually and based on the head of the section's experience and also based on simple rules like FIFO or EDD.Scheduling through the APS software uses internal algorithms that provide a different schedule in relation to the allocation of resources and in relation to the beginnings and endings of tasks compared to what is done in practice.These differences have been shown to improve results in terms of the time needed to complete and, consequently, meet due dates.Thus, it can be concluded that obtaining superior results is not intuitive and would hardly be achieved only based on the experience of the decision maker.The results of the present study corroborate that the use of simple rules such as FIFO and EDD does not necessarily offer the best results in practical and complex situations.
As this is a public service that has notably difficulties in meeting due dates, these results are encouraging as they indicate a possible path to be followed to improve the use of public resources, better meet demand and more effectively comply with service to society.
It can also be said that the user-friendly interface and use of the software, through prior training of users, can make the scheduling of forensic scientists' tasks more efficient at Forensic Science unities.It can also free up more time for the decision maker to solve other day-to-day issues.
Finally, it is noted that the difficulties faced in using the software in this case studied were few and, therefore, did not prevent its proper use in this service and shows that an adaptation or reformulation of the software is not necessary to start your application.

Capacity analysis
The use of the software demonstrated that the capacity of the Regional Technical Section of Criminalistics of Três Corações is exceeded during some periods.The number of reports not scheduled by the software varied according to the rule used, ranging from one to twenty-five not allocated.
The same finding can be made from the actual allocation given by the unit's record history.Some reports took several months to complete and were carried out in parallel with many other reports by the same forensic scientist.

FINAL CONSIDERATIONS
Operations scheduling in Forensic Science services, although notably a difficult task, is usually not carried out with the aid of APS-type systems.However, the complexity and dynamism of this activity suggests precisely the use of software to aid decision making (Enns, 1996).The results of the detailed research in this article provide insights into using an APS software in this situation not yet explored in the literature.
The implementation of the reports in the Preactor® software proved to be feasible and reasonably easy considering the simplification already used by the Technical-Scientific Police Superintendence in relation to the gathering of all types of reports in only 13 groups (according to Annex A) with standards times for each group.
The results show that scheduling and rescheduling from the software's algorithms can bring better results than those currently achieved.These better results are related to the average flow time of the reports (disregarding the reports that was not allocated due to lack of capacity).As developing a system for allocating reports to forensic scientists would take considerable time (Nguyen et al., 2017) and would also require considerable investments, these results weigh in favor of choosing to use ready-made software already available on the market.As the results show that the reports flow times were, on average, shorter than the times in real allocations, it can be said that the allocations suggested by the software can bring benefits to the operation.The benefits would be, firstly, to greater responsiveness of the operation and, secondly, to the increase in the number of reports carried out by the same forensic scientists who are currently in the studied STRC.Thus, as stated by Fuchigami and Rangel (2018), the use of computer systems aimed at operations scheduling can increase the practical application of research in the area.
The biggest difficulty for working with the software during the research, which would probably be similar to the case of practical use in the Regional Technical Section of Criminalistics of Três Corações, is registering the shifts of forensic scientists in specific calendars.Also, in practical use it would be necessary for this registration to be done manually month by month, since the shift schedule is defined by month and changes every month.
The analysis of only one group of historical data is the main limitation of the present work.Thus, studies covering longer time horizons could complement the results found here.Other future research that could add conclusions to the present work would be an evaluation of the same rules using average real times from recent years instead of real times collected from just one historical period.
a) reference times (standards) for carrying out forensic science reports (Annex A); b) shift schedule for the month of March 2015 (Annex B); c) registration of forensic science examinations carried out in 2015.

Figure 1
Figure 1 below summarizes the research steps.

Figure 1
Figure 1 -Research steps (standards) established by the Superintendence of Technical-Scientific Police of the Civilian Police from Minas Gerais state.FSS reports were registered as orders with the same entry dates (in the software defined by the earlier start dates) and the processing times equal to the reference times (in the software defined by the operating time, time per item, in the orders window).All the APS rules available in Preactor® were used to generate the schedules, namely, the APS Forwards rule, the APS Backwards rule, the APS Parallel Loading rule and the APS Preferred Sequence rule.The APS Forwards rule sequences all operations forward, and subsequently, the last operation is blocked and all previous operations are sequenced backwards.The APS Backwards rule sequences operations backwards from the delivery date, and subsequently, the first operation is blocked and subsequent operations are sequenced forward.The APS Paralell Loading rule sequences operations based on simulation considering the availability of resources.

Figure 2 -
Figure 2 -Average flow time of the rules

Table 1
registration of FSS reports in the Preactor® software

Table 2
percentage of similarity between each pair of rules in relation to the allocation of forensic reports to forensic scientist

Table 3
percentage of similarity between each pair of rules in relation to the FSS reports allocation to forensic scientists