Custom Commercial Division
Software Projects - By far the most fundamental change in Climate-Master's engineering department during my employment was the development of software to automate the matching and selection of refrigeration components. Previously, this was done by manually iteration to find the same operating point on various refrigeration component performance charts and tables. This manual design process was very tedious and error prone. By curve fitting all the table and charts of equipment used in the design, this process could be accomplished using a PC replicating the iterative steps a person would take. Later, a more appropriate algorithm (a variation of Newton's Method for multi-variable non-linear systems) was applied to the problem reducing the computing time by at least an order of magnitude. As time permitted, the selection software was enhanced to the point that it provided all the calculations necessary to completely design a refrigeration system. This program was developed by a summer student and myself under the guidance of our supervisor, the engineering manager.
The impact of the refrigeration design software was very significant. Refrigeration design could be done in real time as sales quotes were requested, preventing the sales team from making undersized approximations. Before the design software was in use, mistakes in the sales quotation ultimately would be corrected by engineering department only after the sale was made and was completely at the expense of Climate-Master. Avoiding this situation was of of prime importance, and the refrigeration design software completely resolved the problem. The software also created a lot of consistency in the refrigeration design, and properly documented the design parameters used as an added benefit. While the interface and output was primitive by today's standards, the program did perform all of the complex numerical calculations required by Climate-Master to completely design any refrigeration systems offered in their product line. An example of the output of the refrigeration design program can be found here.
A much smaller but important in house software project that I completed was automating the drawing of refrigeration coils. The program would take a part number and generate the bill of materials, total weight, and create a drawing that manufacturing could use to build the part.
Another interesting problem that I studied in depth was shop floor scheduling. Originally an off the shelf package was used, but proved to be very inadequate in it's ability to dynamically reallocate/reschedule tasks to maintain a desired production schedule. Typically, custom air handling units go through a number of steps in the production process. Each unit requires different amounts of resources at each stage, with some units skipping stages completely. Thus, scheduling the start of unit fabrication is not likely in the same order as the required finish and delivery order. Units entering production in the wrong order would deprive resources from higher priority units and unnecessarily lengthen their production time. Sometimes out of sequence jobs would also cause problems with shop floor space. Low priority units would have to be left temporarily while resources were moved to higher priority units. If this happens too many times, the shop floor becomes clogged with partially finished product, and further delays fabrication of the desired units. When 10 or 15 units are in the production phase, with a manufacturing time measured in weeks, this problem quickly becomes very complex. Examples of the Climate-Master shop floor are shown below with a number of units in various stages of assembly.
In attempting to solve the scheduling problem I first gathered large amounts of historical data and fitted curves to approximate a relation between unit specification and the amount of resources required in the various stages of production. Then, using a program to calculate the required resources of all the work in progress, I manually prepared a schedule of units for the various production stages in an order that appeared correct. As the schedule was completed, errors in the order of units entering production would become apparent and I would rearrange the previous tasks to get the desired output order. By doing this manually I became familiar with the algorithm required to automate the task. Unfortunately the company did not survive long enough to see the scheduling software completed. However, the few schedules generated in this fashion demonstrated accuracy of the technique by predicting late deliveries and misallocated resources, which, to the disbelief of many, ultimately came true.
A software project that I started but never fully completed, was automating the creation of customer proposal packages. With the modules that were finished, customer proposals were entirely in an electronic format thus allowing searches and archiving to be done on a computer. This also added to the appearance of the proposal package, and saved enormous amounts of time preparing the package.
Documentation - In the process of quantifying various aspects of the design and fabrication process, I created forms and documents that described the design process. Since the company had grown over a number of years from a garage operation to a 150 person organization, the importance of documentation was becoming increasingly critical. I learned and used Aldus PageMaker to capture design knowledge in bulletins, and created standard forms to be used during the design process. This was an important step in creating projects that could be reviewed at a latter date without ambiguity. Sales staff also benefited with engineering bulletins that could be given to customers.
Working on software projects taught me serious lessons regarding the strengths and weaknesses of automating the design process. The major gains in productivity and efficiency in the engineering department were very apparent, and demonstrated the power of the PC as an engineering tool. When Climate-Master finally went bankrupt it was this software experience that lead to my decision to return to university and formalize my extreme interests in engineering software with a Computing Science degree.
In general, the variety and challenging nature of the work contributed significantly in developing my passion for engineering, design, and computers.