95112201 [J Bell] Q: Do we talk ( or perhaps when do we talk) in lifecycle terms verses operations?
95112201 {E Dean} A: I believe we should always talk in life cycle terms for all functions of the mission, including the top level function "perform mission." Note that functions are expressed in the form {verb,noun}. The functions of the life cycle of operations are conceptualize operations, evaluate operations, design operations, prototype operations, test operations, produce operations, deploy operations, operate operations, support operations, evolve operations, retire operations, and manage operations.
Each of these functions has a cost. The cost distribution of a function is the distribution of costs of all the various ways of implementing that function. The final cost of the function is the cost of the actual implementation of the function. In other word, the cost distribution converges over time to the accounting noise about a single point.
95112202 [J Bell] Q: Do we talk about operations as a phase of a program ( which can include mini-lifecycle activities for system replacements, new acquisitions, etc ) or just the space equivalent of a manufacturing process?
95112202 {E Dean} A: I prefer to think about operations as a function with an associated implementation which has a cost flow over the time during which the operations life cycle functions are being implemented.
I believe that one of the task of COSTLESS is to deploy these functions through a purpose and means deployment. The result would be a tree of operations functions which have associated cost flows which we need to model. An incomplete strawman deployment for the operation of a spacecraft follows:
At the top level, the purpose is to operate the spacecraft. The means, or how we are going to do that, are at the second level. They are by maintaining the spacecraft, collecting the data, processing the data, and communicating the data. Note that maintain spacecraft has been deployed to a third level. A real mission would generate many levels of deployment.
Note that no mention of implementation has been assumed at this point. We can now deploy the various options implementing each lowest level function. These are abstractions of the system to be implemented. A cost and a set of performance measures can be mapped to each function based upon the implementation. By choosing one of the options for each function we obtain an implementation abstraction of the system. This can be evaluated by the associated cost and performance measures. The set of all feasible system abstractions is the system design space. We can obtain the lowest cost system by choosing that feasible implementation abstraction with the lowest cost. This is a combinatorial optimization problem, which is not easy. Design of experiments can be used to sample the system design space to greatly reduce the number of implementation evaluations necessary to obtain a near minimum cost. The concepts employed here are an extension of those used in value engineering.
Note that, above, we have begun the operations life cycle. We have conceptualized the system and we have evaluated the system in terms of cost. Further, in writing out this scenario, we have conceptualized the conceptualization of the system and we have conceptualized the evaluation of the system. In other words, we have been simultaneously conceptualizing the system and the system to bring forth the system. This is concurrent design which is part of concurrent engineering which is part of integrated product and process development. The deployment of the function above is just a formalized process for a typical functional decomposition from systems engineering.
95112203 [J Bell] Q: Do we include only direct cost or include all indirect cost?
95112203 {E Dean} A: I believe that we should leave the old paradigm of direct and indirect cost behind and learn to think in terms of activity based cost. In a pure form of activity based cost, all activities are direct cost because all cost is allocated to an activity. This would both reduce the accounting noise and increase our ability to estimate costs.
An activity is the implementation of a function by an organization. It is, thus, the natural means of measuring the effort or work of human endeavor which is driven by the purpose described by the function. We use cost as the measure of this purposive effort.
95112204 [J Bell] Q: Do we consider the total runout cost ( which might be the tactic for a single launch type mission) or do we consider on an annual basis for open-ended programs with variable level of activity per annum.?
95112204 {E Dean} A: I prefer to view the estimated cost flows of all functions over the duration of a mission. We, in fact, cannot know the actual cost flows until the mission is complete, and then only within the accounting noise. It is a rare case when we know the mission termination date in advance, thus, a total runout cost make little sense. If I were asked to describe a NASA mission to Congress, I would like to say that we expected the mission development to require about $X with an annual mission cost of about $Y per year for the expected Z year life of the mission.
Note that the annual cost includes the cost to deploy, operate, support, evolve, retire, and manage the system which implements the mission. The annual mission cost of a multiunit system may also include production cost.
The mission cost, the sum of the annual mission costs, as opposed to the mission development cost, is the cost to society described by Taguchi. His quadratic model of cost versus variation from target suggests that the minimum mission cost coincides with perfect accomplishment of the mission. To accomplish that, we must design for perfect accomplishment of the mission. Quality function deployment was designed to accomplish that. It is a process which starts by rigorously defining the mission targets and then establishing an organization to minimize the variation from those mission targets. It has the associated side benefit of reducing the time and cost to develop the system to accomplish the mission.
95112805 [J Bell] Q: How the heck in all this maze do I know when I have printed all the files?
95112805 {E Dean} A: You probably never will because I lost track about six months ago. Anyway, it's time you get out of the linear thinking mode and get nonlinear and recursive. That's the wave of the future. However, since I aim to please my customers, this is an evenings worth of start in that direction. Just for you! However, as soon as you print it, I'm gonna improve it! That is why I have begun to put the content improvement date at the top of the page so you can see that first thing to determine whether you need to print it again or not.
Send questions, answers, observations, inputs, and suggestions for improvement here. All are appreciated.
Originated on 951122 | Improved on 951129
Author Ed Dean |
Curator Al Motley