Patent Application
20080065253
An exemplary embodiment of the method according to the present invention is described in greater detail with reference to the attached drawing. In the drawing, the figures show:
The exemplary embodiment of the present invention described in the following relates to a production process for manufacturing motor vehicles. This production process encompasses the following ten stations through which a production object cycles in succession, in order to manufacture a motor vehicle out of the same:
The vehicle scheduling and final inspection do not require any cycle times, so they are not considered in the following. FIG. I illustrates the order in which the production objects cycle through the other eight stations in the production process.
Each station includes one or more subprocesses. The subprocesses are delimited from each other in such a way that no subprocesses are carried out in parallel or alternatively. Rather, the subprocesses are defined in such a way that any branching thereof occurs only within a subprocess. For example, subprocess 100.3 (“surface finishing”) includes the two work steps 110.1 (““base coat finish”) and 110.2 (“top coat finish”). In work step 110.1, the production objects completed in the body-in-white station undergo a cathodic dip painting, which is followed by the application of a base coat (“filler”). The top coat, which determines the color of the motor vehicle, is subsequently applied in work step 110.2. A clear-coat finish is then added. Depending on the specified color, the top coat and, as a function thereof, the base coat are selected.
In accordance with the present invention, a sequence 70 of production objects 20.1, 20.2, . . . cycles through this production process, from start to finish. In the beginning, the production object exists only “on paper”; at the end of the production process, a completed motor vehicle has been produced. In parallel thereto, a sequence 50 of orders 10.1, 10.2, . . . cycles through the same production process. In this example, each order relates to a motor vehicle. This motor vehicle is manufactured in accordance with order-specific instructions, thus, in such a way that it meets the customer's requirements specified in the order. Typically, the production object does not begin its cycle through the production process until the order is at hand. Each motor vehicle is preferably manufactured on the basis of an order. Each order relates to a buildable motor vehicle, and the execution of each order is at least begun upon receipt of the order. As a result, the exact same number of orders and production objects cycle through the production process.
Additional fictitious orders, which relate to an unfinished motor vehicle, are also preferably generated. A production object is manufactured, for example, on the basis of such a fictitious order, and is then intentionally destroyed in a test or trial during production.
The motor vehicles are preferably manufactured in a fixed-cycle production in the production process. A planned cycle time T is predefined for the entire production process. Two consecutive production objects of production-object sequence 70 are delivered at time interval T to a subprocess.
The order exists in electronic form and includes, for instance, the following specifications for a motor vehicle as a product to be manufactured in accordance with order-specific instructions:
Data records of a database describe the orders of the order sequence. A data record for an order is created upon receipt of the order. The data record remains in the database until a motor vehicle has been completed in accordance with the order, and an invoice has been issued and paid. Another set of data records in the database describes the production-object sequence. The data record of a production object includes the feature characteristics of the production object manufactured up to that point. Once a production object has cycled through subprocess 100.3, the data record includes, inter alia, the following information about the production object:
From the scheduled final inspection date for each order, one derives, on the one hand, the start of production for this order, and, on the other hand, the delivery dates for the supplied subsystems. To this end, a retrospective calculation is made, starting from the final inspection date, as a function of attainable cycle times through subprocesses and in dependence upon available resources.
A so-called order penetration point 300 is defined in the production process. This order penetration point 300 is the point where a binding mutual assignment of order sequence 50 and production-object sequence 70 is made. Thus, from this point on, an order is permanently assigned to each production object of sequence 70. Order penetration point 300 is placed in the production process in such a way that, on the one hand, it occurs as far back in the production process as possible, and, on the other hand, in such a way that many subprocesses, in which subsystems having a wide range of variants and often varying from order to order are installed in the production object, do not occur until after the order penetration point. In this example, the order penetration point has been placed immediately upstream of the interior assembly. The order sequence is altered at the order penetration point if the need arises, such as when the first production object and the first order do not match.
Orders are placed with the suppliers who deliver the subsystems which are installed in subprocesses downstream of order penetration point 300, on the basis of order sequence 50. A supplier may be an external supplier, i.e., a legally independent business entity, or an internal supplier, i.e., a division of the vehicle manufacturer. The production control according to the present invention does not make any distinction between internal and external suppliers. Some subsystems, such as casting molds for cylinder heads, are needed for manufacturing order-specific production objects, but are not installed in them.
Orders for suppliers are derived from each order of order sequence 50 by using a parts list of the motor vehicle. It may be necessary to manufacture a plurality of units of the subsystem for one motor vehicle, such as four seats per motor vehicle. This gives rise to a supply-order sequence for each supplier.
Order penetration point 300 is placed as far back as possible in the production process. In this way, the suppliers are given a longest possible lead time, namely the time between the entry of the production object in first subprocess 100.1 of the production process and the reaching of order penetration point 300. Order penetration point 300 is preferably placed upstream of subprocess 100.5 (“interior assembly station”). Considered collectively, the subsystems produced for the interior assembly, such as cable trees, dashboard assembly, and seats, are so order-specific that they can generally only be used for one single production object.
One selection point, whose function is described in the following, resides upstream of each of the following subprocesses:
A sorting buffer having spaces for production objects is provided at those selection points in which physical production objects are selected. In the example of
At least one additional sorting buffer is preferably provided for subsystems produced on the basis of the order sequence installed downstream of order penetration point 300. A subsystem of this kind is placed in a sorting buffer for subsystems, when the production object into which the subsystem is to be installed, arrives at the installation location later than scheduled.
At each of these selection points, a selection process is repeatedly carried out, in which an order from the order sequence and a production object from the production-object sequence or residing in the sorting buffer, which match one another, are selected. In this comparison, the features of an order are preferably compared to those features of a production object that are produced or modified in the subsequent subprocess, and not to those that remain unchanged in the subprocess. The selected production object is delivered to the subsequent subprocess, where it is processed in accordance with the selected order.
An order is compared to a production object by comparing the data record for the order to the data record for the production object. This comparison is preferably performed fully automatically, without any human intervention.
In the preferred specific embodiment, the selection processes are carried out by an industry-standard master computer used for production control. This master computer used for production control has built-in redundancy and thus minimal downtime. The master computer used for production control has read and write access to the database having the data records for orders and production objects.
As described above, a matching production object and order are selected at every selection point in each selection process. Each of the subprocesses having an upstream selection point is preferably assigned a selection subset of those features that were produced in the preceding subprocesses. A production object and an order are assessed as matching one another when every product feature of the order belonging to the selection subset, is consistent with all of the features of the production object.
As a feature, each selection subset preferably includes the target completion date required by the subprocess, thus, the latest date by when a production object matching the order must have been processed in the subprocess in accordance with the order and have exited the subprocess.
In addition, each subprocess is assigned a processing subset. The features of the processing subset of a subprocess 100.x are processed in subprocess 100.x. A processing order for the subprocess is derived with the aid of the features of a selected order and the features of the processing subset. The production object is derived in the subprocess in accordance with the processing order.
For example, a selection subset having the following features is assigned to subprocess 100.2 (body-in-white station):
The processing subset of subprocess 100.2 includes the following features, for example:
Features, such as the color and type of paint finish or the engine version, which do not yet play a role in subprocess 100.2, but rather first in subsequent subprocesses, are neither included in the selection subset nor in the processing subset of subprocess 100.2.
At selection point 200.2, a production object of a specific model series and of a specific body version is selected for subprocess 100.2. The work order derived is for manufacturing a production object of this model series and this body version including the features “left-hand drive” and “sunroof”.
The selection subset of subprocess 100.3 (surface finishing station) includes the following features, for example:
The processing subset of subprocess 100.3 includes the following features, for example:
At selection point 200.3, for example, a production object of a specific model series and of a specific body version having the features “left-hand drive” and “sunroof” is selected for subprocess 100.5 and for a scheduled final inspection date. The work order derived for subprocess 100.5 stipulates a specific color, as well as the type of paint finish to be used for painting this production object.
The selection subset of subprocess 100.5 (interior installation station) includes the following features, for example:
The processing subset of subprocess 100.5 (interior installation station) includes the following features, for example:
A processing subset is also predefined for subprocess 100.6 (chassis and suspension station). There is no need for a selection subset, because an order is permanently assigned to a production object at order penetration point 300.
During a selection process, the production objects in the sorting buffer are searched through to locate a production object which matches the first order of the order sequence. If a matching production object is found, it is selected together with the first order. The selected production object is removed from the sorting buffer and delivered to the first subprocess.
If no production object matching the first order of the order sequence is found in the sorting buffer, then the first order is compared to the production objects in the production-object sequence, in succession, and, to be precise, beginning with the first production object. It is checked whether free spaces are still available in the sorting buffer for all production objects residing in the production-object sequence upstream of a production object that matches the first order. In the case that sorting buffer has altogether N spaces and N—1 of these spaces are already occupied by production objects, and in the case that N—2 non-matching production objects are located upstream of the first matching production object of the production-object sequence, it is then checked whether N>=N—1+N—2. If this condition is fulfilled, the N—2 non-matching production objects are placed in the sorting buffer, and the first matching production object and the first order are selected.
If, on the other hand, N<N—1+N—2, then it is not possible to move up a production object that matches the first order and to deliver it to the particular subprocess that follows, simply by using the sorting buffer. For such cases, an electronic buffer memory for orders is provided at each selection point. In the example of
It is preferably ensured that an order does not remain in such a buffer memory longer than a predefined maximum dwell time. For that reason, when the dwell time of at least one order in the buffer memory up to that point is greater than or equal to a predefined dwell-time limit, the following steps are implemented during a selection process:
Upstream of this order penetration point 300, an order is assigned to a production object only temporarily, for example, for the particular subprocess that follows; and an order may be assigned in one subprocess to one production object and, in a subsequent subprocess, to another production object. A copy of the order sequence is generated at the selection points upstream of order penetration point 300, thus, in the example of
A sorting buffer for production objects preferably includes N sorting sub-buffers SB—1, . . . , SB_N. A production object is able to be placed in a selected sorting sub-buffer regardless of the fill level and the capacity utilization of the other sorting sub-buffers. All sorting sub-buffers preferably have the same number of spaces for production objects. Each sorting sub-buffer is formed as a lane. A distribution device in the form of at least one cross conveyor, including a storage and retrieval system having a production object placed thereon, traverses between these lanes. This cross conveyor is able to pick up a production object that is to be placed, from the production-object sequence and place the same in the selected lane. In addition, it is able to remove a production object, which has been placed in a lane, from the lane and deliver it to the subprocess.
Using a valuation function, when a production object is to be placed in the sorting buffer, and one of the sorting sub-buffers is to be selected for this purpose, a current valuation is computed for each sorting sub-buffer with respect to the production object. For each sorting sub-buffer SB_k, this valuation function produces a weighted analysis of the n individual criteria C—1(SB_k), . . . , C_n(SB_k), using n predefined weighting factors ω—1, . . . ω—n, where ω—1+ . . . +ω—n=1. Algorithm val(SB_k)=ω—1*C—1(SB_k)+ . . . +ω_n*C_n(SB_k) is used to compute valuation val(SB_k). The sorting sub-buffer receiving the highest valuation is selected.
One alternative specific embodiment provides for initially arranging the sorting sub-buffers in a sequence according to each individual criterion, the sorting sub-buffers being sorted in descending order according to the valuations with respect to this individual criterion. As a result, altogether n sequences are produced. Each sorting sub-buffer thereby receives n place numbers in these n sequences. The first sorting sub-buffer in a sequence receives place number 1, the subsequent one, place number 2, and so forth. The n place numbers of a sorting sub-buffer are subsequently added. That sorting sub-buffer is selected for which the smallest sum of place numbers is obtained. The production object is placed in this selected sorting sub-buffer.
The following individual criteria enter into the valuation function:
When selecting a sorting sub-buffer, it is a goal for as many variants of production objects as possible to reside in each sorting sub-buffer, and at all times. The closer one comes to reaching this goal, the less of an effect a downtime experienced by a sorting sub-buffer has on the processing of the order sequence.
When selecting a sorting sub-buffer, it is preferably first determined which sorting sub-buffers are currently empty, and, among these empty sorting sub-buffers, one is selected, for example the one into which the production object is able to be placed in the least amount of time. Typically, this single criterion is only dependent on the geometry of the sorting buffer and possibly on the production line, but not on the individual production objects or orders.
In the case that a sorting sub-buffer is experiencing downtime due to a breakdown, it is not considered during the selection process until it is available again. It is also possible that some production objects are only able to be placed in a few of the sorting sub-buffers, for example because of their dimensions or because they require a specific ambient temperature. In such a case, it is ascertained before the selection process, which sorting sub-buffers come under consideration at all, for placing the production object.
One refinement provides for modifying the third single criterion in such a way that, the greater the dissimilarity between the production object to be placed and the production objects actually residing in the sorting sub-buffer, the higher the degree of priority is given to a sorting sub-buffer. In the case that the production objects in a sorting sub-buffer are dissimilar to one another and, for that reason, similar production objects are always distributed over different sorting sub-buffers, there is then a greater probability that a production object matching the first order can be removed from the sorting buffer, even when one sorting sub-buffer is experiencing downtime due to a breakdown, making it necessary to revert to other sorting sub-buffers.
The following algorithm is preferably used to determine the value for sorting sub-buffer SB—1 with respect to the third single criterion:
Let PO be the production object to be placed and let PO—1, . . . , PO_m be the m production objects residing in SB—1 before PO is placed. The orders and production objects are compared on the basis of r attributes A(1), . . . , A(r). Examples of attributes are engine versions, left-hand drive or right-hand drive, and the presence or absence of special appointments. Let ω(1), . . . , ω(r) be predefined weighting factors for the r attributes. For the r attributes, r distance measures dist(1), . . . , dist(r) are defined. For s=1, . . . , r, dist(s) defines a measure of the dissimilarity between two possible attribute values a and b of attribute A(s). Frequently, dist(s) (a,b)=1, in the case that a is equal to b, and dist(s)(a,b)=0, in the case that a is unequal to b. Another measure of the dissimilarity is preferably defined for an attribute, for example, which relates to the color of the paint finish of motor vehicles. The greater the deviation between two different paint colors, the greater is the value for dist. Considered here are the outlay entailed in converting a painting line from one color paint to another color paint, and the effects that residual particles of the old color paint have on a paint finish in a new color paint.
Let a(1), . . . , a(r) be the attribute values of production object PO to be placed. For k=1, . . . , m let bk(1), . . . , bk(r) be the attribute values of production object PO_k, which already resides in sorting sub-buffer SB—1. The value with respect to the third single criterion is denoted by dist(PO, SB—1) and is a measure of the dissimilarity of PO from m production objects PO—1, . . . , PO_m in sorting sub-buffer SB—1. Algorithm dist(PO, SB—1)=min {dist (PO, PO—1), . . . , dist (PO, PO_m)} is used to calculate this value. For k=1, . . . , m, dist(PO, PO_k) is a measure of the dissimilarity between PO and PO_k, which is calculated in accordance with algorithm dist(PO, PO_k)=ω(1)*dist(1)(a(1), bk(1)) . . . +ω(r)*dist(r)(a(r), bk(r)).
Orders 10.1, 10.2, 10.3, . . . from customers for vehicles of a specific model series are arranged in an order sequence 50. The manufacturing of production objects for products of this model series is begun on the basis of this order sequence 50. One after another, these production objects exit subprocess 100.2 (body-in-white station) in production-object sequence 20.1, 20.2, 20.3 . . . . A copy 60 of this order sequence 50 is generated, including order copies 10.1, 10.2, 10.3, . . . .
Value 3*T, thus three cycle times, for example, is predefined as dwell-time limit DL for electronic buffer memory 400.3 of selection point 200.3. Thus, a limit of three selection processes is derived. This limit results from the following compromise:
When an order in the buffer memory reaches the dwell-time limit, the attempt is made to find a matching production object for this order. In the case that the first matching production object is not the first of the production-object sequence, then the production objects upstream of the first matching are searched through.
One consideration is that as few orders as possible be processed behind schedule.
The following describes in detail the execution of selection processes.
The selection processes at selection point 200.3 for the current model series begin at point in time T—0 and take place at points in time T_i=T—0+i*T (i=0, 1, 2, 3, . . . ). The time required for executing a selection process is short in comparison to cycle time T.
At point in time T—0, order 10.1 of copy 60 of order sequence 50 and matching production object 20.1 are selected. Production object 20.1 is delivered to subprocess 100.3 and processed in the same in accordance with selected order 10.1. Selected order 10. I has a relative position of 0 in the selection sequence, in comparison to order sequence 50.
In a snapshot view,
At point in time T—1=T—0+T, it is ascertained that order 10.2 and production object 20.2 do not match, because order 10.2 specifies a left-hand drive, whereas production object 20.2 is a right-hand drive vehicle. Consequently, the other production objects in production-object sequence are compared to order 10.2. It is ascertained that production object 20.4 matches order 10.2. Located upstream of production object 20.4 are two production objects which are both able to be placed in sorting buffer 500.3. Therefore, production object 20.4 is selected and moved up to the first position in the production-object sequence. Order 10.2 is likewise selected, and production object 20.4 is processed in accordance with order 10.2 in the subsequent subprocess 100.3. Therefore, after point in time T—1, the two production objects 20.2 and 20.3 reside in sorting buffer 500.3. In a snapshot view,
The next order 10.3 in copy 60 or order sequence 70 is first compared to the two production objects in sorting buffer 500.3 at point in time T—2=T—0+2*T. Order 10.3 does not match production object 20.2, because order 10.3 specifies a motor vehicle without a sun roof, whereas production object 20.3 is a motor vehicle with a sun roof. On the other hand, order 10.3 matches production object 20.2. Therefore, these two are selected. Production object 20.2 is removed from the sorting buffer, delivered to subprocess 100.3, and processed in the same in accordance with order 10.3.
The next order 10.4 in copy 60 is compared to production object 20.2 in sorting buffer 500.3 at point in time T—3=T—0+3*T. The two match and are selected. Production object 20.2 is removed from the sorting buffer.
At point in time T—4=T—0+4*T, the production objects of production-object sequence 70 are searched through to find a production object which matches the next order 10.5 of copy 60. The next production object that matches is production object 20.9. To move this production object up to the first position, four free spaces are needed in sorting buffer 500.3, because upstream of 20.9, four other productions objects 20.5 through 20.8 still come before matching production object 20.9. However, sorting buffer 500.3 only has three free spaces. Therefore, no production object matching order 10.5 can be moved up to the first position, and order 10.5 is placed in buffer memory 400.3. The next order 10.6 of copy 60 is compared to the production objects of production-object sequence 70. The next production object that matches is production object 20.7. The two production objects 20.5 and 20.6 located upstream of this matching production object are placed in sorting buffer 500.3. Production object 20.7 and order 10.6 are selected. Selected order 10.7 [sic. 10.6] has a relative position of +1, because it is delivered to subprocess 100.3 one position earlier than that stipulated by original order sequence 50.
In a snapshot view,
At point in time T—5=T—0+5*T, the orders in buffer memory 400.3 are first compared to production objects in sorting buffer 500.3. The only order 10.5 in the buffer memory had already been compared to production objects 20.5 and 20.6 in sorting buffer 500.3 and been identified as not matching. The first production object of production-object sequence 70 that matches order 10.6, is production object 20.9. This production object can now be moved up to the first position in production-object sequence 70, since, in exchange, only one production object, namely production object 20.8, still needs to be placed in sorting buffer 500.3. Therefore, order 10.6 and production object 20.9 are selected. Production object 20.8 is placed in sorting buffer 500.3 and order 10.6 is removed from buffer memory 400.3. Selected order 10.6 has a relative position of −1.
At selection point in time T—6=T—0+6*T, next order 10.7 is first compared to the three production objects 20.5, 20.6 and 20.8 in sorting buffer 500.3. However, none of the three production objects matches order 10.7. First production object 20.10 of production-object sequence 70 does not match order 10.7 either. Since no more free spaces are available in sorting buffer 500.3, order 10.7 is deferred, in that it is placed in buffer memory 400.3. The next order 10.8 is also first compared to the three production objects in sorting buffer 500.3, however it does not match any of them. On the other hand, order 10.8 matches first production object 20.10 of production-object sequence 70. Therefore, 10.8 and 20.10 are selected. Selected order 10.8 has a relative position of+1.
At selection point in time T—7=T—0+7*T, buffer memory 400.3 is initially searched to find a production object matching order 10.7. The three production objects in sorting buffer 500.3 do not match order 10.7. First production object 20.11 of production-object sequence 70 does not match order 10.7 either. Since no more free spaces are available in sorting buffer 500.3, order 10.7 remains in buffer memory 400.3. The next order 10.9 is also first compared to the three production objects in sorting buffer 500.3, however it does not match any of them. On the other hand, order 10.9 matches first production object 20.11 of production-object sequence 70. Therefore, 10.9 and 20.11 are selected. Selected order 10.9 has a relative position of+1.
In a snapshot view,
At point in time T—8=T—0 +8*T, order 10.7 residing in buffer memory 400.3 is first compared to next production object 20.12, however, they do not match. Therefore, order 10.7 remains in buffer memory 400.3. The next order 10.10 does, in fact, match production object 20.15 of production-object sequence 70, but not the next production object 20.12. Since the matching production object 20.13 cannot be moved up to the first position, order 10.10 is likewise placed in buffer memory 400.3. The next order 10.11 matches the next production object 20.12. Therefore, 10.11 and 20.12 are selected. Selected order 10.11 has a relative position of+2.
At point in time T—9=T—0+9*T, it is ascertained that two orders reside in buffer memory 400.3. Order 10.7 has already resided in buffer memory for two selection processes, order 10.10 not yet for any. Therefore, order 10.7 is first compared to the next production object 20.13, however, the two do not match. On the other hand, order 10.10 matches production object 20.13. For that reason, 10.10 and 20.13 are selected, and order 10.10 is removed from buffer memory 400.3. Selected order 10.10 has a relative position of 0.
In a snapshot view,
At point in time T—10=T—0+10*T, it is ascertained that order 10.7 has already resided in buffer memory 400.3 for three selection processes. Thus, an upper bound is reached, and the order is treated preferentially. Matching production object 20.15 is not able to be moved up to the first position using sorting buffer 500.3, because no free spaces are available. Therefore, in order to continue, one of the following method steps is carried out:
The third alternative is not technically feasible for subprocess 100.3. The fourth alternative would lead to a product in accordance with order 10.7 being completed much later than agreed upon. However, the second alternative is frequently not at all practicable, or is expensive and, therefore, seldom used. The first alternative leads to order 10.7 being processed with an even greater delay relative to the scheduled sequence. However, since only one production object needs to be removed from sorting buffer 500.3 for the first alternative, the first alternative is tried first. Thus, in this situation, not only the next selection process, but also the selection process after the next is scheduled.
It is ascertained that order 10.12 does not match any production object in sorting buffer 500.3. On the other hand, production object 20.5 in sorting buffer 500.3 matches order 10.13. Therefore, production object 20.5 and order 10.13 are selected. Production object 20.5 is removed from sorting buffer 500.3 and delivered to subprocess 100.3. Selected order 10.13 has a relative position of+2. Order 10.12 is placed in buffer memory 400.3.
At point in time T—11=T—0+1*T, the sequence already previously established is executed. Production object 20.14 is placed in the sorting buffer. Order 10.7 and production object 20.15 are selected. Order 10.7 is removed from buffer memory 400.3. Selected order 10.7 has a relative position of−5.
At point in time T—12=T—0+12*T, remaining order 10.12 in buffer memory 400.3 is first compared to the three production objects in sorting buffer 500.3. It is ascertained that order 10.12 matches production object 20.6. Therefore, 10.12 and 20.6 are selected. Order 10.12 is removed from buffer memory 400.3; production object 20.6 from sorting buffer 500.3.
At point in time T—13=T—0+13*T, it is ascertained that order 10.14 matches production object 20.8. Therefore, both are selected, and 20.8 is removed from sorting buffer 500.3. At point in time T—14=T—0+14*T, it is ascertained that order 10.15 matches production object 20.14. Therefore, both are selected, and 20.14 is removed from sorting buffer 500.3. Thus, in this example, copy 60 of order sequence 50 and, thus, also order sequence 50 are processed.
The table in the following illustrates the sequence of the selection processes. The figures, which indicate the situation arrived at in each case, are entered. In this case:
In this example, 4 of the 15 orders were placed in the buffer memory, and 7 of the 15 orders have neither undergone a prefetching nor a postfetching, but are faithful-to-position.
As already described above, a data record in a database is generated for each order and each production object. An industry-standard master computer used for production control has read and write access to this database. It is possible to realize each electronic buffer memory as a separate database and to physically copy data records. Computing time and memory capacity are saved when no data records are copied. Instead, additional data fields are created and modified when implementing the method. This is described in the following.
For one order, each data record includes the following data fields.
The selection processes are advantageously logged at each selection point. In this connection, a log recording the order and production object that are selected is generated each time. Thus, the log generated at a selection point 200.x includes a series of pairs, each including one order and one production object matching the order. The information pertaining to an order includes a unique identifier. The information pertaining to one production object likewise includes a clear identifier, as well as the characteristics of the production object with respect to all of the features which were processed in one of the subprocesses prior to selection point 200.x and, therefore, which belong to the above described processing subset of one of these subprocesses.
A data record for a production object includes the following data fields:
When an order is “waiting” for a subprocess, thus when a production object had been processed in a preceding subprocess on the basis of the order, and the production object had exited from the preceding subprocess, but had not yet been selected for the next subprocess, that subprocess is noted for which the order is waiting. The data field for an electronic buffer memory is, of course, only filled when an order of the copy had been placed in the electronic buffer memory. After the order is removed from the electronic buffer memory, the data field for the buffer memory is emptied.
In the example of
When carrying out a selection process, the master computer used for production control searches through the data records for orders and, in each case, searches for a matching production object and order. If they are found, an identifier for the selected production object is noted in the data record for the selected order. Conversely, an identifier for the selected order is noted in the data record for the selected production object. The data fields “actual position” of the order data record and “position” of the production object are filled with the current values. The copy of the order sequence is generated in that the data fields “actual position” of the order data records are filled and modified. As soon as an order reaches “order penetration point” 300, the values in “actual position” and “setpoint position” are identical, so that only the value of “setpoint position” is still needed.
The data fields “actual position”, “actual points in time”, “electronic buffer memory” and “production object of the order data records”, as well as “position” and “order” of the production-object data records are preferably emptied at regular intervals and described using the current values. These current values are ascertained beforehand. In this manner, a defined starting point is created at regular intervals. For example, preventive maintenance on the entire production process is performed each night. During these maintenance procedures, the just mentioned data fields are emptied and filled with the ascertained current values.
By automatically evaluating the log generated at selection point 200.x, the positional quality described in the following, of the preceding subprocess is able to be ascertained.
By automatically evaluating the log from selection point 200.3, the positional quality of subprocess 100.2 is able to be ascertained, for example. The sequence of the orders in the order sequence prior to subprocess 100.2 is compared to the sequence subsequent to subprocess 100.2 In the preferred specific embodiment, a copy 60 of order sequence 50 is generated at selection point 200.3. Instead of the original order sequence, only the copy is altered by the selection processes at selection point 200.3. For that reason, after the selection processes are carried out, the copy is compared to the original.
In the example illustrated by
In this example, the greatest degree of prefetching, that is the largest relative position, amounts to+2 positions. Because, in this example, the sorting buffer can only accommodate three production objects, an upper bound of three is set on the prefetching operation. The greatest degree of postfetching, that is the amount of the smallest relative position, amounts to five positions.
A histogram can be used to illustrate the positional quality that is attainable using sorting buffer 500.3.
The generated logs described above for a reference type of production object are used for sizing the sorting buffer for a type to be produced later. This is described on the basis of the example of selection point 200.3 using sorting buffer 500.3. A log is used including original order sequence 50 and sequence 70 of production objects in the sequence in which the production objects exited preceding subprocess 100.2. The logged original sequence functions as a reference sequence of reference orders. Logged sequence 70 of the production objects functions as an electronic copy of the reference production-object sequence.
In this example, the following possible values are predefined for the maximum number of available spaces of sorting buffer 500.3 for the production objects: 0, 3, 6 and 9 spaces. An electronically available model of sorting buffer 500.3 is generated. This model can be set to a predefined maximum number of available spaces.
A simulation is performed for each one of the predefined possible values. In this simulation, the model of sorting buffer 500.3 and electronic buffer memory 400.3 are used.
The simulation is performed for three storage spaces precisely in the manner described above. The following table clarifies the simulation for zero storage spaces, thus for the case that no sorting buffer at all is provided. In this simulation, the following strategy is used in the selection processes:
In this example, 14 of the 15 orders were placed in the buffer memory, and only one of the orders has neither undergone a prefetching nor a postfetching operation, but is faithful-to-position. As a secondary result, the simulation reveals the quality of subprocess 100.2 without the use of sorting buffer 500.3, because, inter alia, the maximum postfetching and prefetching, as well as the average faithfulness-to-position are ascertained.
For each of the predefined possible values, the following quality parameters are determined by the simulations:
An operating point is selected on one of these functions. A number of available spaces is defined by the selection. Preferably, a lower bound of, for example, 75% is preset on the sequence efficiency, or an upper bound of, for example, 25 relative positions on the maximum postfetching. Because acquiring and operating the sorting buffer is all the more expensive, and because all the more space the sorting buffer takes up, the more spaces it has, it is designed to have as few spaces as possible. When a lower bound on the sequence efficiency is preset, an operating point is preferably selected to be as close as possible to the lower bound. Correspondingly, when an upper bound on the maximum postfetching is preset, an operating point is selected to be as close as possible to the upper bound. One alternative method provides for selecting an operating point for which the slope of the function is approximately 45 degrees or −45 degrees. This operating point is selected because it yields a good compromise between the requirement for an excellent sequence efficiency or a low maximum postfetching, and the requirement for as few as possible spaces.
In
In the specific embodiment described up to now, the sequence of reference orders and the sequence of reference production objects remain unchanged when used in the simulation. A variation of this procedure allows for the possibility that the reference type differs from the product type whose production is to be improved using the sorting buffer to be sized. This distinction is considered by altering the two reference sequences before a simulation is performed for each one of the possible values of spaces. At least one of the following modifications is made:
A further refinement of this method makes it possible to compare different ranges of variants under the given the product type. For example, two functions are generated: one for a range of variants of 16 different top coat colors, another for a range of variants of 24 different top coat colors. Two functions of the sequence quality are generated, and two functions of the maximum prefetching are generated, in each case as a function of the number of available spaces. Represented graphically, this shows a family of curves. Accordingly, it is possible to ascertain what effects an improvement in the positional quality in subprocess 100.2 will have on the sequence efficiency attained by sorting buffer 500.3. Two functions of the sequence efficiency and two functions of the maximum prefetching are generated, namely one in the case of an unchanged positional quality and one in the case of an improved positional quality.
| Number | Date | Country | Kind |
|---|---|---|---|
| 0321768.4 | Sep 2003 | GB | national |
| Filing Document | Filing Date | Country | Kind | 371c Date |
|---|---|---|---|---|
| PCT/EP04/01732 | 2/21/2004 | WO | 00 | 2/13/2007 |
