Differences in the modeling of planned and actual operations

In the previous article “Why do we need to model individual and typical scenarios?” I told about the need to model both typical and individual scenarios in cases where the customer wants to automate three functions:

1. Modeling operations and scenarios based on typical schemes.
2. Comparison of the results with the planned.
3. Adjustment plans.

Now I will focus on the difference in modeling the upcoming work and the accomplished actions.



To describe the accomplished actions we need four coordinates: three spatial and one time. Of course, it’s quite expensive to model this four-dimensional space in the forehead - too much computer memory will be needed, and analyzing the accumulated data will become impossible. To reduce the model to an acceptable level, we are able to isolate some subspaces from this four-dimensional space, endowing them with certain properties. For example, if we are talking about a stool, then we, as a rule, neglect the fact that this stool changes with time - we consider it a static object. This allows us to describe its geometry once, in order to refer to it later. If we are talking about a drop of water, then, indicating its trajectory, we, as a rule, neglect the fact that it evaporated during the flight. If there are many drops: so many that it is not possible to describe the trajectory of each of them, we go further and talk about the trajectories of the drops as a whole, introducing a new object into consideration - the flow of water, and, in particular, the fountain. We say that the fountain “beats”, but we consider it already as a static object. There are few such tricks in modeling and all of them are classified.
')
When describing the future, four coordinates are no longer enough. The future is multivariate. This means that the concept of planned tolerances must be introduced into the model of planned operations. If, in describing the past, we describe the trajectory of the system, then in describing the future, we have to model many such trajectories with possible tolerances. How does it look in practice?

Case 1

Let it be necessary to calculate the cost of the order. We know that any picker from the picking department can do this work. Suppose in the department of five pickers. When planning work, we do not know which employee will do the job. Therefore, we write: the operation “calculate the cost of order No. 123” will be performed by any order picker from the picking department.

Let order calculation completed. Now we know that the calculation was made by calculator number four - Sidorov Ivan Ivanovich. To simulate this fact, we write: the operation “calculating the cost of order No. 123” was performed by the “calculator No. 4”, which was performed by “Sidorov Ivan Ivanovich”

Case 2

Let the planned operation to assemble the furniture. The plan states: the beginning of work on July 22 from 9-00 to 18-00. So we modeled the variability of the future. We do not know when exactly the start of the furniture assembly operation will occur. Therefore, we had to specify a valid range of values.

Let the furniture assembly operation begin. We know that it began at 17-45. This fact is reflected in the model as an exact value of the date of the beginning of the operation.

Case 3

Let it be known that if the cost of contract No. 123 is less than 1,000,000 rubles, then the sales manager will be the supervisor of the execution of the contract. If the cost of contract No. 123 turns out to be more than 1,000,000 rubles, then the head of the sales department should become the curator. So we model the future variability through branching operators: If (condition) Then (action).

Let the real value of the contract was 998,000 rubles. Therefore, the curator was appointed sales manager number 2 - Ivanov. We model it this way: since the contract value is less than 1,000,000 rubles, the sales manager No. 2 was appointed curator.

Modeling past and future

Not always the past is known reliably. If we know the vulgar with some accuracy, modeling it looks the same as modeling a variable future. But the meaning of these models is different.

Case 1

Suppose we do not know who specifically made the calculation of the order, but we know that some kind of accountant from the picking department. Then we simulate this knowledge in this way: the operation “calculating the cost of order No. 123” was performed by one of the calculators of the picking department.

Case 2

Suppose we know that the furniture assembly operation took place on July 22nd, it is also known that the beginning of its implementation took place from 9-00 to 18-00. To simulate this fact, we say: the beginning of the operation took place in the interval between 9-00 and 18-00.

Case 3

Suppose that we do not know who was the curator of the contract: whether the sales manager number 2, or the head of the department. In addition, we do not know the exact amount of the contract. However, we know that the sales manager No. 2 was the curator if the amount of the transaction was less than a million rubles, and the head of the sales department if the amount of the transaction was more than a million rubles. In modeling, we say: if the contract value was less than a million rubles, then the sales manager was its supervisor, if more, the department head.

The treatment of models of the past and the future

Thus, the same structures can be used to model the past and the future, but the interpretation of data in these structures will be different. If, by modeling the future, we are modeling its variability, then by modeling the past, we are modeling the accuracy of our knowledge about it. Speaking about the future, we model the probability of occurrence of certain events (Markov chains). Modeling the past, we simulate the reliability of certain data.

Operation modeling

Modeling the future, we actually solve two problems. The first task is the description of the goal achievement algorithm. When solving this problem, we invent an algorithm for solving the problem (planning from above). The second task is to find sufficient resources for this algorithm (planning from below). As a result, the problem is solved as follows:

1. A hypothesis is advanced - an algorithm.
2. For this algorithm all necessary resources are indicated.
3. If resources are not found, then we put forward another algorithm for solving the problem as a working hypothesis and repeat the cycle.

At the enterprise, the solution of the problem and the fit of this solution to the existing resources are decided by a technologist. The technologist develops a way to solve the problem, based on the known capabilities of the contractor. The resulting solution - there is a work plan, the trajectory of movement.

Performing operations

Further, this plan begins to be executed. Execution of real operations brings the unit to some state. This state may be in the zone of permissible deviations from the planned trajectory. This means that the movement was performed according to plan. If the state goes beyond the tolerances, then this will require an adjustment of the trajectory of achieving the goal.

findings

Thus, the production performs the following functions:

1. Creating problem solving hypotheses - trajectories.
2. Test hypotheses on existing resources.
3. Rejection of non-working hypotheses and acceptance of working hypotheses for execution.
4. Determination of intermediate rescheduling points.
5. Creating tasks to the nearest points of planning.
6. Collection of information on the actual work performed and the determination of the actual state at the time of completion of tasks.
7. Comparison of the received state with the planned one.
8. Correction of trajectories in the direction of travel.

Different methods of planning and management use different methods for assigning tasks and rescheduling points.

When working on shift-daily tasks, the planning horizon is a day. This means that analysis and rescheduling is done once a day.

When working according to standard scenarios in BPMN notation, the planning horizon is the time of the current operation. That is, after the completion of the current operation, a decision is made on further actions: to start or not to start the next operation. Unfortunately, there is no possibility in BPMN to reschedule the work execution scenario. This leads to the fact that analysts are forced to invent complex structures (case management) , which “plug the hole” in the existing methodology.

When working with a Gantt chart, the adjustment of plans occurs upon the occurrence of control events: the beginning or end of operations, the occurrence of milestones. This is the most thoughtful methodology for planning and monitoring operations.

At the same time, if the planned operation requires resources for modeling a multivariate future, then the operation that has been completed requires less resources for its modeling.

Next time I will talk about modeling activities.